Highlights

Box plot

Added a new plot type: box plot. It's useful to visualise a distribution when you have a set of values for each sample.

from multiqc.plots import box
self.add_section(
    ...,
    plot=box.plot(
        {
            "sample 1": [4506, 4326, 3137, 1563, 1730, 3254, 2259, 3670, 2719, ...],
            "sample 2": [2145, 2011, 3368, 2132, 1673, 1993, 6635, 1635, 4984, ...],
            "sample 3": [1560, 1845, 3247, 1701, 2829, 2775, 3179, 1724, 1828, ...],
        },
        pconfig={
            "title": "Iso-Seq: Insert Length",
        },
    )
)

Note the difference with the violin plot: the box plot visualises the distributions of many values within one sample, whereas the violin plot shows the distribution of one metric across many samples.

pyproject.toml

The setup.py file has been superseded by pyproject.toml for the build configuration.
Note that now for new modules, an entry point should be added to pyproject.toml instead of setup.py, e.g.:

[project.entry-points."multiqc.modules.v1"]
afterqc = "multiqc.modules.afterqc:MultiqcModule"

Heatmap

The heatmap plot now supports passing a dict as input data, and also supports a zlab
parameter to set the label for the z-axis:

from multiqc.plots import heatmap
self.add_section(
    ...,
    plot=heatmap.plot(
        {
            "sample 1": {"sample 2": 0, "sample 3": 1},
            "sample 2": {"sample 1": 0, "sample 3": 0},
            "sample 3": {"sample 1": 1, "sample 2": 0, "sample 3": 1},
        },
        pconfig={
            "title": "Sample comparison",
            "zlab": "Match",
        },
    )
)

MultiQC updates

• New plot type: box plot (#2358)
• Add "Export to CSV" button for tables (#2394)
• Replace setup.py with pyproject.toml (#2353)
• Heatmap: allow a dict dicts of data (#2386)
• Heatmap: add zlab config parameter. Show xlab, ylab, zlab in tooltip (#2387)
• Warn if run_modules contains a non-existent module (#2322)
• Catch non-hashable values (dicts, lists) passed as a table cell value (#2348)
• Always create JSON even when MegaQC upload is disabled (#2330)
• Use generic font family for Plotly (#2368)
• Use a padded span with nowrap instead of   before suffixes in table cells (#2395)
• Refactor: fix unescaped regex strings (#2384)

Fixes:

• Pin the required Plotly version and add a runtime version check (#2325)
• Bar plot: preserve the sample order (#2339)
• Bar plot: fix inner gap in group mode (#2321)
• Violin: filter Inf values (#2380)
• Table: Fix use of the no_violin (ex-no_beeswarm) table config flag (#2376)
• Heatmap: prevent from parsing numerical sample names (#2349)
• Work around call of full_figure_for_development to avoid Kaleido errors (#2359)
• Auto-generate plot id when pconfig=None (#2337)
• Fix: infinite dmax or dmin fail JSON dump load in JavaScript (#2354)
• Fix: dump pconfig for MegaQC (#2344)

New modules

• IsoSeq
  • Iso-Seq contains the newest tools to identify transcripts in PacBio single-molecule sequencing data (HiFi reads). cluster and refine commands are supported.
• Space Ranger
  • Works with data from 10X Genomics Visium. Processes sequencing reads and images created using
    the 10x Visium platform to generate count matrices with spatial information.
  • New MultiQC module parses Space Ranger quality reports.

Module updates

• bcl2fastq: fix the top undetermined barcodes plot (#2340)
• DRAGEN: add few coverage metrics in general stats (#2341)
• DRAGEN: fix showing the number of found samples (#2347)
• DRAGEN: support gvcf_metrics (#2327)
• fastp: fix detection of JSON files (#2334)
• HTSeq Count: robust file reading loop, ignore .parquet files (#2364)
• Illumina InterOp Statistics: do not set 'scale': False as a default (#2350)
• mosdepth: fix regression in showing general stats (#2346)
• Picard: Crosscheck Fingerprints updates (#2388)
  • add a heatmap for LOD scores besides a table
  • if too many pairs in table, skip those with Expected status
  • use the warn status for Inconclusive
  • add a separate sample-wise table instead of general stats
  • sort tables by status, not by sample name
  • add a column "Best match" and "Best match LOD" in tables
  • hide the LOD Threshold column
• PURPLE: support v4.0.1 output without version column (#2366)
• Samtools: support new coverage command (#2356)
• UMI-tools: support new extract command (#2296)
• Whatshap: make robust when a stdout is appended to TSV (#2361)

New Contributors

• @grst made their first contribution in https://github.com/MultiQC/MultiQC/pull/1945

Full Changelog: https://github.com/MultiQC/MultiQC/compare/v1.20...v1.21

Highlights

New plotting library

MultiQC v1.20 comes with totally new plotting code for MultiQC reports. This is a huge change to the report output. We've done our best to maintain feature parity with the previous plotting code, but please do let us know if you spot any bugs or changes in behaviour by creating a GitHub issue.

This change comes with many improvements and new features, and paves the way for more in the future. To find out more, read the associated blog post.

For now, you can revert to the previous plotting code by using the highcharts report template (multiqc --template highcharts). This will be removed in v1.21.

Note that there are several plotting configuration options which have been removed:

• click_func
• cursor
• tt_percentages (use tt_suffix: "%")
• Bar plot:
  • use_legend (automatically hidden if there is only 1 category)
• Line plot:
  • labelSize
  • xDecimals, yDecimals (automatic if all values can be cast to int)
  • xLabelFormat, yLabelFormat (use tt_label)
  • pointFormat
• Heatmap:
  • datalabel_colour
  • borderWidth

Moved GitHub and docker repositories

The v1.20 release is also the first release we've had since we moved the MultiQC repositories. Please note that the code is now at MultiQC/MultiQC (formerly ewels/MultiQC) and the same for the Docker repository. The GitHub repo should automatically redirect, but it's still good to update any references you may have.

MultiQC updates

• Support Plotly as a new backend for plots (#2079)
  • The default template now uses Plotly for all plots
  • Added a new plot type violin (replaces beeswarm)
  • Moved legacy Highcharts/Matplotlib code under an optional template highcharts
    (#2292)
• Move GitHub repository to MultiQC organisation (#2243)
• Update all GitHub actions to their latest versions (#2242)
• Update docs to work with Astro 4 (#2256)
• Remove unused dependency on future library (#2258)
• Fix incorrect scale IDs caught by linting (#2272)
• Docs: fix missing v prefix in docker image tags (#2273)
• Unicode file reading errors: attempt to skip non-unicode characters (#2275)
• Heatmap: check if value is numeric when calculating min and max (#2276)
• Add filesearch_file_shared config option, remove unnecessary per-module shared flags in search patterns (#2227)
• Use alternative method to walk directory using pathlib (#2277)
• Export config.output_dir in MegaQC JSON (#2287)
• Drop support for module tags (#2278)
• Pin Pillow package, wrap add_logo in try-except (#2312)
• Custom content: support multiple datasets (#2291)
• Configuration: fix reading config.output_fn_name and --filename (#2314)

New modules

• Bamdst (#2161)
  • Bamdst is a lightweight tool to stat the depth coverage of target regions of bam file(s).
• MetaPhlAn (#2262)
  • MetaPhlAn is a computational tool for profiling the composition of microbial communities from metagenomic shotgun sequencing data.
• MEGAHIT (#2222)
  • MEGAHIT is an ultra-fast and memory-efficient NGS assembler
• Nonpareil (#2215)
  • Estimate metagenomic coverage and sequence diversity.

Module updates

• Bcftools: order variant depths plot categories (#2289)
• Bcftools: add missing self.ignore_samples in stats (#2288)
• BCL Convert: add index, project names to sample statistics and calculate mean quality for lane statistics. (#2261)
• BCL Convert: fix duplicated yield for 3.9.3+ when the yield is provided explicitly in Quality_Metrics (#2253)
• BCL Convert: handle samples with zero yield (#2297)
• Bismark: fix old link in Bismark docs (#2252)
• Bismark: fix old link in docs (#2252)
• Cutadapt: support JSON format (#2281)
• HiFiasm: account for lines with no asterisk (#2268)
• HUMID: add cluster statistics (#2265)
• mosdepth: add additional summaries to general stats #2257 (#2257)
• Picard: fix using multiple times in report: do not pass module.anchor to self.find_log_files (#2255)
• QualiMap: address NBSP as thousands separators (#2282)
• Seqera Platform CLI: updates for v0.9.2 (#2248)
• Seqera Platform CLI: handle failed tasks (#2286)

New Contributors

• @a-detiste made their first contribution in https://github.com/MultiQC/MultiQC/pull/2258
• @mbeavitt made their first contribution in https://github.com/MultiQC/MultiQC/pull/2268
• @almahans made their first contribution in https://github.com/MultiQC/MultiQC/pull/2262
• @kmlhpk made their first contribution in https://github.com/MultiQC/MultiQC/pull/2273
• @studioph made their first contribution in https://github.com/MultiQC/MultiQC/pull/2277
• @vsmalladi made their first contribution in https://github.com/MultiQC/MultiQC/pull/1503

Full Changelog: https://github.com/MultiQC/MultiQC/compare/v1.19...v1.20

Highlights

An early Christmas present 🎁 Happy holidays everyone! 🎄

This release is mostly bugfixes and minor additions, whilst we lay the groundwork for some bigger updates coming in the new year. Still, there are plenty of goodies in here. Enjoy!

See the full changes in this release here: https://github.com/ewels/MultiQC/compare/v1.18...v1.19

MultiQC updates

• Add missing table id in DRAGEN modules, and require id in plot configs in strict mode (#2228)
• Config table_columns_visible and table_columns_name: support flat config and table_id as a group (#2191)
• Add sort_samples: false config option for bar graphs (#2210)
• Upgrade the jQuery tablesorter plugin to v2 (#1666)
• Refactor pre-Python-3.6 code, prefer f-strings over .format() calls (#2224)
• Allow specifying default sort columns for tables with defaultsort (#1667)
• Create CODE_OF_CONDUCT.md (#2195)
• Add .cram to sample name cleaning defaults (#2209)

MultiQC bug fixes

• Re-add run into the multiqc namespace (#2202)
• Fix the "square": True flag to scatter plot to actually make the plot square (#2189)
• Fix running with the --no-report flag (#2212)
• Fix guessing custom content plot type: do not assume first row of a bar plot data are sample names (#2208)
• Fix detection of changed specific module in Changelog CI (#2234)

Module updates

• BCLConvert: fix mean quality, fix count-per-lane bar plot (#2197)
• deepTools: handle missing data in plotProfile (#2229)
• Fastp: search content instead of file name (#2213)
• GATK: square the BaseRecalibrator scatter plot (#2189)
• HiC-Pro: add missing search patterns and better handling of missing data (#2233)
• Kraken: fix UnboundLocalError (#2230)
• Kraken: fixed column keys in genstats (#2205)
• QualiMap: fix BamQC for global-only stats (#2207)
• Picard: add more search patterns for MarkDuplicates, including MarkDuplicatesSpark (#2226)
• Salmon: add library_types, compatible_fragment_ratio, strand_mapping_bias to the general stats table (#1485)

New Contributors

• @gsmashd made their first contribution in https://github.com/ewels/MultiQC/pull/2205
• @kclem made their first contribution in https://github.com/ewels/MultiQC/pull/2235

Full Changelog: https://github.com/ewels/MultiQC/compare/v1.18...v1.19

Highlights

Better configs

As of this release, you can now set all of your config variables via environment variables! (see docs).

Better still, YAML config files can now use string interpolation to parse environment variables within strings (see docs), eg:

report_header_info:
  - Contact E-mail: !ENV "${NAME:info}@${DOMAIN:example.com}"

Picard refactoring

In this release, there was a significant refactoring of the Picard module.
It has been generalized for better code sharing with other Picard-based software, like Sentieon and Parabricks.
As a result of this, the standalone Sentieon module was removed: Sentieon QC files will be interpreted directly as Picard QC files.

If you were using the Sentieon module in your pipelines, make sure to update any places that reference the module name:

• MultiQC command line (e.g. replace --module sentieon with --module picard).
• MultiQC configs (e.g. replace sentieon with picard in options like run_modules, exclude_modules, module_order).
• Downstream code that relies on names of the files in multiqc_data or multiqc_plots saves (e.g., multiqc_data/multiqc_sentieon_AlignmentSummaryMetrics.txt becomes multiqc_data/multiqc_picard_AlignmentSummaryMetrics.txt).
• Code that parses data files like multiqc_data/multiqc_data.json.
• Custom plugins and templates that rely on HTML anchors (e.g. #sentieon_aligned_reads becomes #picard_AlignmentSummaryMetrics).
• Also, note that Picard fetches sample names from the commands it finds inside the QC headers (e.g. # net.sf.picard.analysis.CollectMultipleMetrics INPUT=Szabo_160930_SN583_0215_AC9H20ACXX.bam ... -> Szabo_160930_SN583_0215_AC9H20ACXX), whereas the removed Sentieon module prioritized the QC file names. To revert to the old Sentieon approach, use the use_filename_as_sample_name config flag.

MultiQC updates

• Config can be set with environment variables, including env var interpolation (#2178)
• Try find config in ~/.config or $XDG_CONFIG_HOME (#2183)
• Better sample name cleaning with pairs of input filenames (#2181)
• Software versions: allow any string as a version tag (#2166)
• Table columns with non-numeric values and now trigger a linting error if scale is set (#2176)
• Stricter config variable typing (#2178)
• Remove position:absolute CSS from table values (#2169)
• Fix column sorting in exported TSV files from a matplotlib linegraph plot (#2143)
• Fix custom anchors for kraken (#2170)
• Fix logging spillover bug (#2174)

New Modules

• Seqera Platform CLI (#2151)
  • Seqera Platform CLI reports statistics generated by the Seqera Platform CLI.
• Xenome (#1860)
  • A tool for classifying reads from xenograft sources.
• xengsort (#2168)
  • xengsort is a fast xenograft read sorter based on space-efficient k-mer hashing

Module updates

• fastp: add version parsing (#2159)
• fastp: correctly parse sample name from --in1/--in2 in bash command. Prefer file name if not fastp.json; fallback to file name when error (#2139)
• Kaiju: fix division by zero error (#2179)
• Nanostat: account for both tab and spaces in v1.41+ search pattern (#2155)
• Pangolin: update for v4: add QC Note , update tool versions columns (#2157)
• Picard: Generalize to directly support Sentieon and Parabricks outputs (#2110)
• Sentieon: Removed the module in favour of directly supporting parsing by the Picard module (#2110)
  • Note that any code that relies on the module name needs to be updated, e.g. -m sentieon will no longer work
  • The exported plot and data files will be now be prefixed as picard instead of sentieon, etc.
  • Note that the Sentieon module used to fetch the sample names from the file names by default, and now it follows the Picard module's logic, and prioritizes the commands recorded in the logs. To override, use the use_filename_as_sample_name config flag

The one with the new logo

Highlights:

• Introducing the new MultiQC logo!
• Adding support for Python 3.12 and dropping support for Python 3.7
• New --require-logs to fail if expected tool outputs are not found
• Rename --lint to --strict
• Modules should now use ModuleNotFoundError instead of UserWarning when no logs are found
• 2 new modules and updates to 9 modules.

MultiQC updates

• Add CI action changelog.yml to populate the changelog from PR titles, triggered by a comment @multiqc-bot changelog (#2025, #2102, #2115)
• Add GitHub Actions bot workflow to fix code linting from a PR comment (#2082)
• Use custom exception type instead of UserWarning when no samples are found. (#2049)
• Lint modules for missing self.add_software_version (#2081)
• Strict mode: rename config.lint to config.strict, crash early on module or template error. Add MULTIQC_STRICT=1 (#2101)
• Matplotlib line plots now respect xLog: True and yLog: True in config (#1632)
• Fix matplotlib linegraph and bargraph for the case when xmax < xmin in config (#2124)
• Add --require-logs flag to error out if requested modules not used (#2109)
• Fixes for python 3.12
  • Replace removed distutils (#2113)
  • Bundle lzstring (#2119)
• Drop Python 3.6 and 3.7 support, add 3.12 (#2121)
• Just run CI on the oldest + newest supported Python versions (#2074)
• New logo
• Set name and anchor for the custom content "module" #2131
• Fix use of shutil.copytree when overriding existing template files in tmp_dir (#2133)

New Modules

• Bracken
  • A highly accurate statistical method that computes the abundance of species in DNA sequences from a metagenomics sample.
• Truvari (#1751)
  • Truvari is a toolkit for benchmarking, merging, and annotating structural variants

Module updates

• Dragen: make sure all inputs are recorded in multiqc_sources.txt (#2128)
• Cellranger: Count submodule updated to parse Antibody Capture summary (#2118)
• fastp: parse unescaped sample names with white spaces (#2108)
• FastQC: Add top overrepresented sequences table (#2075)
• HiCPro: Fix parsing scientific notation in hicpro-ashic. Thanks @Just-Roma (#2126)
• HTSeq Count: allow counts files with more than 2 columns (#2129)
• mosdepth: fix prioritizing region over global information (#2106)
• Picard: Adapt WgsMetrics to parabricks bammetrics outputs (#2127)
• Picard: MarkDuplicates: Fix parsing mixed strings/numbers, account for missing trailing 0 (#2083, #2094)
• Samtools: Add MQ0 reads to the Percent Mapped barplot in Stats submodule (#2123)
• WhatsHap: Process truncated input with no ALL chromosome (#2095)

Highlight: Reporting software versions

New in v1.16 - software version information can now automatically parsed from log output where available, and added to MultiQC in a standardised manner. It's shown in the MultiQC report next to section headings and in a dedicated report section, as well as being saved to multiqc_data. Where version information is not available in logs, it can be submitted manually by using a new special file type with filename pattern *_mqc_versions.yml. There's the option of representing groups of versions, useful for a tool that uses sub-tools, or pipelines that want to report version numbers per analysis step.

There are a handful of new config scopes to control behaviour: software_versions, skip_versions_section, disable_version_detection, versions_table_group_header.
See the documentation for more (writing modules, supplying stand-alone)

Huge thanks to @pontushojer for the contribution (#1927). This idea goes way back to issue #290, made in 2016! 🎉

MultiQC updates

• Removed simplejson unused dependency (#1973)
• Give config custom_plot_config priority over column-specific settings set by modules
• When exporting plots, make a more clear error message for unsupported FastQC dot plot (#1976)
• Fixed parsing of plot_type: "html" data in json custom content
• Replace deprecated pkg_resources
• Fix the module groups configuration for modules where the namespace is passed explicitly to general_stats_addcols. Namespace is now always appended to the module name in the general stats (2037).
• Do not call sys.exit() in the multiqc.run() function, to avoid breaking interactive environments. #2055
• Fixed the DOI exports in multiqc_data to include more than just the MultiQC paper (#2058)
• Fix table column color scaling then there are negative numbers (1869)
• Export plots as static images and data in a ZIP archive. Fixes the issue when only 10 plots maximum were downloaded due to the browser limitation.

New Modules

• Bakta
  • Rapid and standardized annotation of bacterial genomes, MAGs & plasmids.
• mapDamage
  • mapDamage2 is a computational framework written in Python and R, which tracks and quantifies DNA damage patterns among ancient DNA sequencing reads generated by Next-Generation Sequencing platforms.
• Sourmash
  • Quickly search, compare, and analyze genomic and metagenomic data sets.

Module updates

• BcfTools
  • Stats: fix parsing multi-sample logs (#2052)
• Custom content
  • Don't convert sample IDs to floats (#1883)
• DRAGEN
  • Make DRAGEN module use fn_clean_exts instead of hardcoded file names. Fixes working with arbitrary file names (#1994)
• FastQC:
  • fix UnicodeDecodeError when parsing fastqc_data.txt: try latin-1 or fail gracefully (#2024)
• Kaiju:
  • Fix UnboundLocalError on outputs when Kanju was run with the -e flag (#2023)
• Kraken
  • Parametrize top-N through config (#2060)
  • Fix bug where ranks incorrectly assigned to tabs (#1766).
• Mosdepth
  • Add X/Y relative coverage plot, analogous to the one in samtools-idxstats (#1978)
  • Added the perchrom_fraction_cutoff option into the config to help avoid clutter in contig-level plots
  • Fix a bug happening when both region and global coverage histograms for a sample are available (i.e. when mosdepth was run with --by, see mosdepth docs). In this case, data was effectively merged. Instead, summarise it separately and add a separate report section for the region-based coverage data.
  • Do not fail when all input samples have no coverage (#2005).
• NanoStat
  • Support new format (#1997).
• RSeQC
  • Fix max() arg is an empty sequence error (#1985)
  • Fix division by zero on all-zero input (#2040)
• Samtools
  • Stats: fix "Percent Mapped" plot when samtools was run with read filtering (#1972)
• Qualimap
  • BamQC: Include % On Target in General Stats table (#2019)
• WhatsHap
  • Bugfix: ensure that TSV is only split on tab character. Allows sample names with spaces (#1981)

New Contributors

• @gdrosos made their first contribution in https://github.com/ewels/MultiQC/pull/1973
• @antass made their first contribution in https://github.com/ewels/MultiQC/pull/1991
• @lindenb made their first contribution in https://github.com/ewels/MultiQC/pull/1972
• @odcambc made their first contribution in https://github.com/ewels/MultiQC/pull/1968
• @ehenrion made their first contribution in https://github.com/ewels/MultiQC/pull/1855
• @louperelo made their first contribution in https://github.com/ewels/MultiQC/pull/1959
• @Joon-Klaps made their first contribution in https://github.com/ewels/MultiQC/pull/1903
• @sstrong99 made their first contribution in https://github.com/ewels/MultiQC/pull/1891
• @taylorreiter made their first contribution in https://github.com/ewels/MultiQC/pull/1812
• @FerriolCalvet made their first contribution in https://github.com/ewels/MultiQC/pull/2020

Full Changelog: https://github.com/ewels/MultiQC/compare/v1.15...v1.16

Big speedup in file search step

This release of MultiQC introduces speed improvements to the file search (~5-7x fold improvement) 🎉

One way it does this is by limiting the number of lines loaded by each search pattern. For the vast majority of users, this should have no effect except faster searches. However, in some edge cases it may introduce some breaking changes. Hypothetically, for example:

• If you concatenate log files from multiple tools
• If you have a custom plugin module that we haven't tested

See the troubleshooting docs for more information.

MultiQC updates

• Refactor file search for performance improvements (#1904)
• Bump log_filesize_limit default (to skip large files in the search) from 10MB to 50MB.
• Table code now tolerates lambda function calls with bad data (#1739)
• Beeswarm plot now saves data to multiqc_data, same as tables (#1861)
• Don't print DOI in module if it's set to an empty string.
• Don't sort table headers alphabetically if we don't have an OrderedDict - regular dicts are fine in Py3 (#1866)
• New back-end to preview + deploy the new website when the docs are edited.
• Fixed a lot of broken links in the documentation from the new website change in structure.

New Modules

• Librarian
  • A tool to predict the sequencing library type from the base composition of a supplied FastQ file.

Module updates

• Cell Ranger
  • Bugfix: avoid multiple KeyError exceptions when parsing Cell Ranger 7.x web_summary.html (#1853, #1871)
• DRAGEN
  • Restored functionality to show target BED coverage metrics (#1844)
  • Update filename pattern in RNA quant metrics (#1958)
• filtlong
  • Handle reports from locales that use . as a thousands separator (#1843)
• HUMID
  • Fix bug that prevent HUMID stats files from being parsed (#1856)
• Mosdepth
  • Fix data not written to mosdepth_cumcov_dist.txt and mosdepth_cov_dist.txt (#1868)
  • Update documentation with new file {prefix}.mosdepth.summary.txt (#1868)
  • Fill in missing values for general stats table (#1868)
  • Include mosdepth/summary file paths in multiqc_sources.txt (#1868)
  • Enable log switch for Coverage per contig plot (#1868)
  • Fix y-axis scaling for Coverage distribution plot (#1868)
  • Handle case of intermediate missing coverage x-values in the *_dist.txt file causing a distorted Coverage distribution plot (#1960)
• Picard
  • WgsMetrics: Fix wrong column label (#1888)
  • HsMetrics: Add missing field descriptions (#1928)
• Porechop
  • Don't render bar graphs if no samples had any adapters trimmed (#1850)
  • Added report section listing samples that had no adapters trimmed
• RSeQC
  • Fix ZeroDivisionError error for bam_stat results when there are 0 reads (#1735)
• UMI-tools
  • Fix bug that broke the module with paired-end data (#1845)

MultiQC new features

• Rewrote the Dockerfile to build multi-arch images (amd64 + arm), run through a non-privileged user and build tools for non precompiled python binaries (#1541, #1541)
• Add a new lint test to check that colour scale names are valid (#1835)
• Update github actions to run tests on a single module if it is the only file affected by the PR (#915)
• Add CI testing for Python 3.10 and 3.11
• Optimize line-graph generation to remove an n^2 loop (#1668)
• Parsing output file column headers is much faster.

MultiQC code cleanup

• Remove Python 2-3 compatability from __future__ imports
• Remove unused #!/usr/bin/env python hashbangs from module files
• Add new code formatting tool isort to standardise the order and formatting of Python module imports
• Add Pycln pre-commit hook to remove unused imports

MultiQC updates

• Bugfix: Make config.data_format work again (#1722)
• Bump minimum version of Jinja2 to >=3.0.0 (#1642)
• Disable search progress bar if running with --quiet or --no-ansi (#1638)
• Allow path filters without full paths by trying to prefix analysis dir when filtering (#1308)
• Fix sorting of table columns with text values
• Don't crash if a barplot is given an empty list of categories (#1540)
• New logos! MultiQC is now developed and maintained at Seqera Labs. Updated logos and email addresses accordingly.

New Modules

• Anglerfish
  • A tool designed to assess pool balancing, contamination and insert sizes of Illumina library dry runs on Oxford Nanopore data.
• BBDuk
  • Combines most common data-quality-related trimming, filtering, and masking operations via kmers into a single high-performance tool.
• Cell Ranger
  • Works with data from 10X Genomics Chromium. Processes Chromium single cell data to align reads, generate feature-barcode matrices, perform clustering and other secondary analysis, and more.
  • New MultiQC module parses Cell Ranger quality reports from VDJ and count analysis
• DIAMOND
  • A high-throughput program for aligning DNA reads or protein sequences against a protein reference database.
• DRAGEN-FastQC
  • Illumina Bio-IT Platform that uses FPGA for accelerated primary and secondary analysis
  • Finally merged the epic 2.5-year-old pull request, with 3.5k new lines of code.
  • Please report any bugs you find!
• Filtlong
  • A tool for filtering long reads by quality.
• GoPeaks
  • GoPeaks is used to call peaks in CUT&TAG/CUT&RUN datasets.
• HiFiasm
  • A haplotype-resolved assembler for accurate Hifi reads
• HUMID
  • HUMID is a tool to quickly and easily remove duplicate reads from FastQ files, with or without UMIs.
• mOTUs
  • Microbial profiling through marker gene (MG)-based operational taxonomic units (mOTUs)
• Nextclade
  • Tool that assigns clades to SARS-CoV-2 samples
• Porechop
  • A tool for finding and removing adapters from Oxford Nanopore reads
• PRINSEQ++
  • PRINSEQ++ is a C++ of prinseq-lite.pl program for filtering, reformating or trimming genomic and metagenomic sequence data.
• UMI-tools
  • Work with Unique Molecular Identifiers (UMIs) / Random Molecular Tags (RMTs) and single cell RNA-Seq cell barcodes.

Module updates

• Bcftools stats
  • Bugfix: Do not show empty bcftools stats variant depth plots (#1777)
  • Bugfix: Avoid exception when PSC nMissing column is not present (#1832)
• BclConvert
  • Handle single-end read data correctly when setting cluster length instead of always assuming paired-end reads (#1697)
  • Handle different R1 and R2 read-lengths correctly instead of assuming they are the same (#1774)
  • Handle single-index paired-end data correctly
  • Added a config option to enable the creation of barplots with undetermined barcodes (create_unknown_barcode_barplots with False as default) (#1709)
• BUSCO
  • Update BUSCO pass/warning/fail scheme to be more clear for users
• Bustools
  • Show median reads per barcode statistic
• Custom content
  • Create a report even if there's only Custom Content General Stats there
  • Attempt to cooerce line / scatter x-axes into floats so as not to lose labels (#1242)
  • Multi-sample line-graph TSV files that have no sample name in row 1 column 1 now use row 1 as x-axis labels (#1242)
• fastp
  • Add total read count (after filtering) to general stats table (#1744)
  • Don't crash for invalid JSON files (#1652)
• FastQC
  • Report median read-length for fastqc in addition to mean (#1745)
• Kaiju
  • Don't crash if we don't have any data for the top-5 barplot (#1540)
• Kallisto
  • Fix ZeroDivisionError when a sample has zero reads (#1746)
• Kraken
  • Fix duplicate heatmap to account for missing taxons (#1779)
  • Make heatmap full width
  • Handle empty kreports gracefully (#1637)
  • Fix regex error with very large numbers of unclassified reads (#1639)
• malt
  • Fixed division by 0 in malt module (#1683)
• miRTop
  • Avoid KeyError - don't assume all fields present in logs (#1778)
• Mosdepth
  • Don't pad the General Stats table with zeros for missing data (#1810)
• Picard
  • HsMetrics: Allow custom columns in General Stats too, with HsMetrics_genstats_table_cols and HsMetrics_genstats_table_cols_hidden
• Qualimap
  • Added additional columns in general stats for BamQC results that displays region on-target stats if region bed has been supplied (hidden by default) (#1798)
  • Bugfix: Remove General Stats rows for filtered samples (#1780)
• RSeQC
  • Update geneBody_coverage to plot normalized coverages using a similar formula to that used by RSeQC itself (#1792)
• Sambamba Markdup
  • Catch zero division in sambamba markdup (#1654)
• Samtools
  • Added additional column for flagstat that displays percentage of mapped reads in a bam (hidden by default) (#1733)
• VEP
  • Don't crash with ValueError if there are zero variants (#1681)

MultiQC updates

• Major spruce of the command line help, using the new rich-click package
• Drop some of the Python 2k compatability code (eg. custom requirements)
• Improvements for running MultiQC in a Python environment, such as a Jupyter Notebook or script
  • Fixed bug raised when removing logging file handlers between calls that arose when configuring the root logger with dictConfig (#1643)
• Added new config option custom_table_header_config to override any config for any table header
• Fixed edge-case bug in custom content where a description that doesn't terminate in . gave duplicate section descriptions.
• Tidied the verbose log to remove some very noisy statements and add summaries for skipped files in the search
• Add timezone to time in reports
• Add nix flake support
• Added automatic tweet about new releases

Module updates

• AdapterRemoval
  • Finally merge a fix for counts of reads that are discarded/collapsed (#1647)
• VEP
  • Fixed bug when General Statistics have a value of - (#1656)
• Custom content
  • Only set id for custom content when id not set by metadata (#1629)
  • Fix bug where module wouldn't run if all content was within a MultiQC config file (#1686)
  • Fix crash when info isn't set (#1688)
• Nanostat
  • Removed HTML escaping of special characters in the log to fix bug in jinja2 v3.10 removing jinja2.escape() (#1659)
  • Fix bug where module would crash if input does not contain quality scores (#1717)
• Pangolin
  • Updated module to handle outputs from Pangolin v4 (#1660)
• Somalier
  • Handle zero mean X depth in Sex plot (#1670)
• Fastp
  • Include low complexity and too long reads in filtering bar chart
• miRTop
  • Fix module crashing when ref_miRNA_sum is missing in file. (#1712)
  • Fix module crashing due to zero division (#1719)
• FastQC
  • Fixed error when parsing duplicate ratio when there is nan values in the report. (#1725)

Version 1.12 of MultiQC brings with it a modest number of new modules, a few new core features and a swathe of bugfixes and general improvements. I hope that everyone continues to find it useful! You can see the full changes in this release here: https://github.com/ewels/MultiQC/compare/v1.11...v1.12

Special thanks to the 13 people who had their first MultiQC contributions in this release:

• @HofLucien made their first contribution in https://github.com/ewels/MultiQC/pull/1486
• @jchorl made their first contribution in https://github.com/ewels/MultiQC/pull/1578
• @bjohnnyd made their first contribution in https://github.com/ewels/MultiQC/pull/1489
• @schorlton made their first contribution in https://github.com/ewels/MultiQC/pull/1567
• @MatthiasZepper made their first contribution in https://github.com/ewels/MultiQC/pull/1584
• @g-pacheco made their first contribution in https://github.com/ewels/MultiQC/pull/1587
• @paulstretenowich made their first contribution in https://github.com/ewels/MultiQC/pull/1605
• @yanick made their first contribution in https://github.com/ewels/MultiQC/pull/1595
• @MillironX made their first contribution in https://github.com/ewels/MultiQC/pull/1594
• @sguizard made their first contribution in https://github.com/ewels/MultiQC/pull/1593
• @maleasy made their first contribution in https://github.com/ewels/MultiQC/pull/1552
• @TomaszSuchan made their first contribution in https://github.com/ewels/MultiQC/pull/1271
• @massiddamt made their first contribution in https://github.com/ewels/MultiQC/pull/1021

✨ MultiQC - new features

• Added option to customise default plot height in plot config (#1432)
• Added --no-report flag to skip report generation (#1462)
• Added support for priting tool DOI in report sections (#1177)
• Added support for --custom-css-file / config.custom_css_files option to include custom CSS in the final report (#1573)
• New plot config option labelSize to customise font size for axis labels in flat MatPlotLib charts (#1576)
• Added support for customising table column names (#1255)

🔨 MultiQC - updates

• MultiQC now skips modules for which no files were found - gives a small performance boost (#1463)
• Improvements for running MultiQC in a Python environment, such as a Jupyter Notebook or script
  • Fixed logger bugs when calling multiqc.run multiple times by removing logging file handlers between calls (#1141)
  • Init/reset global state between runs (#1596)
• Added commonly missing functions to several modules (#1468)
• Wrote new script to check for the above function calls that should be in every module (.github/workflows/code_checks.py), runs on GitHub actions CI
• Make table Conditional Formatting work at table level as well as column level. (#761)
• CSS Improvements to make printed reports more attractive / readable (#1579)
• Fixed a problem with numeric filenames (#1606)
• Fixed nasty bug where line charts with a categorical x-axis would take categories from the last sample only (#1568)
• Ignore any files called multiqc_data.json (#1598)
• Check that the config path_filters is a list, convert to list if a string is supplied (#1539)

🎁 New Modules

• CheckQC
  • A program designed to check a set of quality criteria against an Illumina runfolder
• pbmarkdup
  • Mark duplicate reads from PacBio sequencing of an amplified library
• WhatsHap
  • WhatsHap is a software for phasing genomic variants using DNA sequencing reads

🌟 Module feature additions

• BBMap
  • Added handling for qchist output (#1021)
• bcftools
  • Added a plot with samplewise number of sites, Ts/Tv, number of singletons and sequencing depth (#1087)
• Mosdepth
  • Added mean coverage #1566
• NanoStat
  • Recognize FASTA and FastQ report flavors (#1547)

🐛 Module updates

• BBMap
  • Correctly handle adapter stats files with additional columns (#1556)
• bclconvert
  • Handle change in output format in v3.9.3 with new Quality_Metrics.csv file (#1563)
• bowtie
  • Minor update to handle new log wording in bowtie v1.3.0 (#1615)
• CCS
  • Tolerate compound IDs generated by pbcromwell ccs in the general statistics (#1486)
  • Fix report parsing. Update test on attributes ids (#1583)
• Custom content
  • Fixed module failing when writing data to file if there is a / in the section name (#1515)
  • Use filename for section header in files with no headers (#1550)
  • Sort custom content bargraph data by default (#1412)
  • Always save custom content data to file with a name reflecting the section name. (#1194)
• DRAGEN
  • Fixed bug in sample name regular expression (#1537)
• Fastp
  • Fixed % pass filter statistics (#1574)
• FastQC
  • Fixed several bugs occuring when FastQC sections are skipped (#1488, #1533)
  • Clarify general statistics table header for length
• goleft/indexcov
  • Fix ZeroDivisionError if no bins are found (#1586)
• HiCPro
  • Better handling of errors when expected data keys are not found (#1366)
• Lima
  • Move samples that have been renamed using --replace-names into the General Statistics table (#1483)
• miRTrace
  • Replace hardcoded RGB colours with Hex to avoid errors with newer versions of matplotlib (#1263)
• Mosdepth
  • Fixed issue #1568
  • Fixed a bug when reporting per contig coverage
• Picard
  • Update ExtractIlluminaBarcodes to recognise more log patterns in newer versions of Picard (#1611)
• Qualimap
  • Fix ZeroDivisionError in QM_RNASeq and skip genomic origins plot if no aligned reads are found (#1492)
• QUAST
  • Clarify general statistics table header for length
• RSeQC
  • Fixed minor bug in new TIN parsing where the sample name was not being correctly cleaned (#1484)
  • Fixed bug in the junction_saturation submodule (#1582)
  • Fixed bug where empty files caused tin submodule to crash (#1604)
  • Fix bug in read_distribution for samples with zero tags (#1571)
• Sambamba
  • Fixed issue with a change in the format of output from sambamba markdup 0.8.1 (#1617)
• Skewer
  • Fix ZeroDivisionError if no available reads are found (#1622)
• Somalier
  • Plot scaled X depth instead of mean for Sex plot (#1546)
• VEP
  • Handle table cells containing - instead of numbers (#1597)

A summer release for MultiQC 🥳 🏖️ Many thanks to everyone who has contributed!

MultiQC new features

• New interactive slider controls for controlling heatmap colour scales (#1427)
• Added new --replace-names / config sample_names_replace option to replace sample names during report generation
• Added use_filename_as_sample_name config option / --fn_as_s_name command line flag (#949, #890, #864, #998, #1390)
  • Forces modules to use the log filename for the sample identifier, even if the module usually takes this from the file contents
  • Required a change to the clean_s_name() function arguments. All core MultiQC modules updated to reflect this.
  • Should be backwards compatible for custom modules. To adopt new behaviour, supply f instead of f["root"] as the second argument.
  • See the documenation for details: Using log filenames as sample names and Custom sample names.

MultiQC updates

• Make the module crash tracebacks much prettier using rich
• Refine the cli log output a little (nicely formatted header line + drop the [INFO])
• Added docs describing tools for downstream analysis of MultiQC outputs.
• Added CI tests for Python 3.9, pinned networkx package to >=2.5.1 (#1413)
• Added patterns to config.fn_ignore_paths to avoid error with parsing installation dir / singularity cache (#1416)
• Print a log message when flat-image plots are used due to sample size surpassing plots_flat_numseries config (#1254)
• Fix the mqc_colours util function to lighten colours even when passing categorical or single-length lists.
• Bugfix for Custom Content, using YAML configuration (eg. section headers) for images should now work

New Modules

• BclConvert
  • Tool that converts / demultiplexes Illumina Binary Base Call (BCL) files to FASTQ files
• Bustools
  • Tools for working with BUS files
• ccs
  • Generate highly accurate single-molecule consensus reads from PacBio data
• GffCompare
  • GffCompare can annotate and estimate accuracy of one or more GFF files compared with a reference annotation.
• Lima
  • The PacBio Barcode Demultiplexer
• NanoStat
  • Calculate various statistics from a long read sequencing datasets
• ODGI
  • Optimized dynamic genome/graph implementation
• Pangolin
  • Added MultiQC support for Pangolin, the tool that determines SARS-CoV-2 lineages
• Sambamba Markdup
  • Added MultiQC module to add duplicate rate calculated by Sambamba Markdup.
• Snippy
  • Rapid haploid variant calling and core genome alignment.
• VEP
  • Added MultiQC module to add summary statistics of Ensembl VEP annotations.
  • Handle error from missing variants in VEP stats file. (#1446)

Module feature additions

• Cutadapt
  • Added support for linked adapters #1329]
  • Parse whether trimming was 5' or 3' for Lengths of Trimmed Sequences plot where possible
• Mosdepth
  • Include or exclude contigs based on patterns for coverage-per-contig plots
• Picard
  • Add support for CollectIlluminaBasecallingMetrics, CollectIlluminaLaneMetrics, ExtractIlluminaBarcodes and MarkIlluminaAdapters (#1336)
  • New insertsize_xmax configuration option to limit the plotted maximum insert size for InsertSizeMetrics
• Qualimap
  • Added new percentage coverage plot in QM_RNASeq (#1258)
• RSeQC
  • Added a long-requested submodule to support showing the TIN (Transcript Integrity Number) (#737)

Module updates

• biscuit
  • Duplicate Rate and Cytosine Retention tables are now bargraphs.
  • Refactor code to only calculate alignment statistics once.
  • Fixed bug where cytosine retentions values would not be properly read if in scientific notation.
• bcl2fastq
  • Added sample name cleaning so that prepending directories with the -d flag works properly.
• Cutadapt
  • Plot filtered reads even when no filtering category is found (#1328)
  • Don't take the last command line string for the sample name if it looks like a command-line flag (#949)
• Dragen
  • Handled MultiQC crashing when run on single-end output from Dragen (#1374)
• fastp
  • Handle a ZeroDivisionError if there are zero reads (#1444)
• FastQC
  • Added check for if overrepresented_sequences is missing from reports (#1281)
• Flexbar
  • Fixed bug where reports with 0 reads would crash MultiQC (#1407)
• Kraken
  • Handle a ZeroDivisionError if there are zero reads (#1440)
  • Updated search patterns to handle edge case (#1428)
• Mosdepth
  • Show barplot instead of line graph for coverage-per-contig plot if there is only one contig.
• Picard
  • RnaSeqMetrics - fix assignment barplot labels to say bases instead of reads (#1408)
  • CrosscheckFingerprints - fix bug where LOD threshold was not detected when invoked with "new" picard cli style. fixed formatting bug (#1414)
  • Made checker for comma as decimal separator in HsMetrics more robust (#1296)
• qc3C
  • Updated module to not fail on older field names.
• Qualimap
  • Fixed wrong units in tool tip label (#1258)
• QUAST
  • Fixed typo causing wrong number of contigs being displayed (#1442)
• Sentieon
  • Handled ZeroDivisionError when input files have zero reads (#1420)
• RSEM
  • Handled ZeroDivisionError when input files have zero reads (#1040)
• RSeQC
  • Fixed double counting of some categories in read_distribution bar graph. (#1457)

This is a relatively small release focussing just on bug fixes 🐛 - the last release revealed a couple of nasty ones which I felt were bad enough to justify a fast patch.. (99 bugs - fix 1, now you've got 103 bugs)

Many thanks to everyone who reports these problems along with example data 🕵🏻

MultiQC updates

• Allow scientific notation numbers in colour scheme generation
  • Fixed bug with very small minimum numbers that only revelead itself after a bugfix done in the v1.10 release
• Require at least rich version 9.4.0 to avoid SpinnerColumn AttributeError (#1393)
• Dropped the Skipping search pattern log message from a warning to debug
• Moved directory prepending with -d back to before sample name cleaning (as it was before v1.7) (#1264)
• If linegraph plot data goes above ymax, only discard the data if the line doesn't come back again (#1257)
• Allow top_modules to be specified as empty dicts (#1274)
• Properly ignore .snakemake folders as intended (#1395)

Module updates

• bcftools
  • Fixed bug where QUAL value . would crash MultiQC (#1400)
• bowtie2
  • Fix bug where HiSAT2 paired-end bar plots were missing unaligned reads (#1230)
• Deeptools
  • Handle plotProfile data where no upstream / downstream regions have been calculated around genes (#1317)
  • Fix IndexError caused by mysterious -1 in code.. (#1275)
• FastQC
  • Replace NaN with 0 in the Per Base Sequence Content plot to avoid crashing the plot (#1246)
• Picard
  • Fixed bug in ValidateSamFile module where additional whitespace at the end of the file would cause MultiQC to crash (#1397)
• Somalier
  • Fixed bug where using sample name cleaning in a config would trigger a KeyError (#1234)

Many thanks to everyone's patience in waiting for this release, it is much appreciated!

Update for developers: Code linting

This is a big change for MultiQC developers. I have added automated code formatting and code linting (style checks) to MultiQC. This helps to keep the MultiQC code base consistent despite having many contributors and helps me to review pull-requests without having to consider whitespace. Specifically, MultiQC now uses three main tools:

• Black - Python Code
• Prettier - Everything else (almost)
• markdownlint-cli - Stricter markdown rules

All developers must run these tools when submitting changes via Pull-Requests! Automated CI tests now run with GitHub actions to check that all files pass the above tests. If any files do not, that test will fail giving a red ❌ next to the pull request.

For further information, please see the documentation.

MultiQC updates

New MultiQC Features

• --sample-filters now also accepts show_re and hide_re in addition to show and hide. The _re options use regex, while the "normal" options use globbing.
• MultiQC config files now work with .yml file extension as well as .yaml
  • .yaml will take preference if both found.
• Section comments can now also be added for General Statistics
  • section_comments: { general_stats: "My comment" }
• New table header config option bgcols allows background colours for table cells with categorical data.
• New table header config options cond_formatting_rules and cond_formatting_colours
  • Comparable functionality to user config options table_cond_formatting_rules and table_cond_formatting_colours,
    allowes module developers to format table cell values as labels.
• New CI test looks for git merge markers in files
• Beautiful new progress bar from the amazing willmcgugan/rich package.
• Added a bunch of new default sample name trimming suffixes (see 8ac5c7b)
• Added timeout-minutes: 10 to the CI test workflow to check that changes aren't negatively affecting run time too much.
• New table header option bars_zero_centrepoint to treat 0 as zero width bars and plot bar length based on absolute values

New Modules

• EigenStratDatabaseTools
  • Added MultiQC module to report SNP coverages from eigenstrat_snp_coverage.py in the general stats table.
• HOPS
  • Post-alignment ancient DNA analysis tool for MALT
• JCVI
  • Computes statistics on genome annotation.
• ngsderive
  • Forensic analysis tool useful in backwards computing information from next-generation sequencing data.
• OptiType
  • Precision HLA typing from next-generation sequencing data
• PURPLE
  • A purity, ploidy and copy number estimator for whole genome tumor data
• Pychopper
  • Identify, orient and trim full length Nanopore cDNA reads
• qc3C
  • Reference-free QC of Hi-C sequencing data
• Sentieon
  • Submodules added to catch Picard-based QC metrics files

Module updates

• DRAGEN
  • Fix issue where missing out fields could crash the module (#1223)
  • Added support for whole-exome / targetted data (#1290)
• featureCounts
  • Add support for output from Rsubread (#1022)
• fgbio
  • Fix ErrorRateByReadPosition to calculate ymax not just on the overall error_rate, but also specific base errors (ex. a_to_c_error_rate, a_to_g_error_rate, ...). (#1215)
  • Fix ErrorRateByReadPosition plotted line names to no longer concatenate multiple read identifiers and no longer have off-by-one read numbering (ex. Sample1_R2_R3 -> Sample1_R2) ([#1304)
• Fastp
  • Fixed description for duplication rate (pre-filtering, not post) ([#1313)
• GATK
  • Add support for the creation of a "Reported vs Empirical Quality" graph to the Base Recalibration module.
• hap.py
  • Updated module to plot both SNP and INDEL stats (#1241)
• indexcov
  • Fixed bug when making the PED file plots (#1265)
• interop
  • Added the % Occupied metric to Read Metrics per Lane table which is reported for NovaSeq and iSeq platforms.
• Kaiju
  • Fixed bug affecting inputs with taxa levels other than Phylum (#1217)
  • Rework barplot, add top 5 taxons (#1219)
• Kraken
  • Fix ZeroDivisionError (#1276)
  • Add distinct minimizer heatmap for KrakenUniq style duplication information (#1333)
• MALT
  • Fix y-axis labelling in bargraphs
• MACS2
  • Add number of peaks to the General Statistics table.
• mosdepth
  • Enable prepending of directory to sample names
  • Display contig names in Coverage per contig plot tooltip
• Picard
  • Fix HsMetrics bait percentage columns (#1212)
  • Fix ConvertSequencingArtifactToOxoG files not being found (#1310)
  • Make WgsMetrics histogram smoothed if more than 1000 data points (avoids huge plots that crash the browser)
  • Multiple new config options for WgsMetrics to customise coverage histogram and speed up MultiQC with very high coverage files.
  • Add additional datasets to Picard Alignment Summary (#1293)
  • Add support for CrosscheckFingerprints (#1327)
• PycoQC
  • Log10 x-axis for Read Length plot (#1214)
• Rockhopper
  • Fix issue with parsing genome names in Rockhopper summary files (#1333)
  • Fix issue properly parsing multiple samples within a single Rockhopper summary file
• Salmon
  • Only try to generate a plot for fragment length if the data was found.
• verifyBamID
  • Fix CHIP value detection (#1316).

New Custom Content features

• General Stats custom content now gives a log message
• If id is not set in JSON or YAML files, it defaults to the sample name instead of just custom_content
• Data from JSON or YAML now has data keys (sample names) run through the clean_s_name() function to apply sample name cleanup
• Fixed minor bug which caused custom content YAML files with a string data type to not be parsed

Bug Fixes

• Disable preservation of timestamps / modes when copying temp report files, to help issues with network shares (#1333)
• Fixed MatPlotLib warning: FixedFormatter should only be used together with FixedLocator
• Fixed long-standing min/max bug with shared minimum values for table columns using shared_key
• Made table colour schemes work with negative numbers (don't strip - from values when making scheme)

Another massive release - many thanks to all of the contributors! Keep those pull-requests and issues coming!

Dropped official support for Python 2

Python 2 had its official sunset date
on January 1st 2020, meaning that it will no longer be developed by the Python community.
Part of the python.org statement reads:

That means that we will not improve it anymore after that day,
even if someone finds a security problem in it.
You should upgrade to Python 3 as soon as you can.

Very many Python packages no longer support Python 2
and it whilst the MultiQC code is currently compatible with both Python 2 and Python 3,
it is increasingly difficult to maintain compatibility with the dependency packages it
uses, such as MatPlotLib, numpy and more.

As of MultiQC version 1.9, Python 2 is no longer officially supported.
Automatic CI tests will no longer run with Python 2 and Python 2 specific workarounds
are no longer guaranteed.

Whilst it may be possible to continue using MultiQC with Python 2 for a short time by
pinning dependencies, MultiQC compatibility for Python 2 will now slowly drift and start
to break. If you haven't already, you need to switch to Python 3 now.

New MultiQC Features

• Now using GitHub Actions for all CI testing
  • Dropped Travis and AppVeyor, everything is now just on GitHub
  • Still testing on both Linux and Windows, with multiple versions of Python
  • CI tests should now run automatically for anyone who forks the MultiQC repository
• Linting with --lint now checks line graphs as well as bar graphs
• New gathered template with no tool name sections (#1119)
• Added --sample-filters option to add show/hide buttons at the top of the report (#1125)
  • Buttons control the report toolbox Show/Hide tool, filtering your samples
  • Allows reports to be pre-configured based on a supplied list of sample names at report-generation time.
• Line graphs can now have Log10 buttons (same functionality as bar graphs)
• Importing and running multiqc in a script is now a little Better
  • multiqc.run now returns the report and config as well as the exit code. This means that you can explore the MultiQC run time a little in the Python environment.
  • Much more refactoring is needed to make MultiQC as useful in Python scripts as it could be. Watch this space.
• If a custom module anchor is set using module_order, it's now used a bit more:
  • Prefixed to module section IDs
  • Appended to files saved in multiqc_data
  • Should help to prevent duplicates requiring -1 suffixes when running a module multiple times
• New heatmap plot config options xcats_samples and ycats_samples
  • If set to False, the report toolbox options (highlight, rename, show/hide) do not affect that axis.
  • Means that the Show only matching samples report toolbox option works on FastQC Status Checks, for example (#1172)
• Report header time and analysis paths can now be hidden
  • New config options show_analysis_paths and show_analysis_time (#1113)
• New search pattern key skip: true to skip specific searches when modules look for a lot of different files (eg. Picard).
• New --profile-runtime command line option (config.profile_runtime) to give analysis of how long the report takes to be generated
  • Plots of the file search results and durations are added to the end of the MultiQC report as a special module called Run Time
  • A summary of the time taken for the major stages of MultiQC execution are printed to the command line log.
• New table config option only_defined_headers
  • Defaults to true, set to false to also show any data columns that are not defined as headers
  • Useful as allows table-wide defaults to be set with column-specific overrides
• New module key allowed for config.extra_fn_clean_exts and config.fn_clean_exts
  • Means you can limit the action of a sample name cleaning pattern to specific MultiQC modules (#905)

New Custom Content features

• Improve support for HTML files - now just end your HTML filename with _mqc.html
  • Native handling of HTML snippets as files, no MultiQC config or YAML file required.
  • Also with embedded custom content configuration at the start of the file as a HTML comment.
• Add ability to group custom-content files into report sections
  • Use the new parent_id, parent_name and parent_description config keys to group content together like a regular module (#1008)
• Custom Content files can now be configured using custom_data, without giving search patterns or data
  • Allows you to set descriptions and nicer titles for images and other 'blunt' data types in reports (#1026)
  • Allows configuration of custom content separately from files themselves (tsv, csv, txt formats) (#1205)

New Modules:

• DRAGEN
  • Illumina Bio-IT Platform that uses FPGA for secondary NGS analysis
• iVar
  • Added support for iVar: a computational package that contains functions broadly useful for viral amplicon-based sequencing.
• Kaiju
  • Fast and sensitive taxonomic classification for metagenomics
• Kraken
  • K-mer matching tool for taxonomic classification. Module plots bargraph of counts for top-5 hits across each taxa rank. General stats summary.
• MALT
  • Megan Alignment Tool: Metagenomics alignment tool.
• miRTop
  • Command line tool to annotate miRNAs with a standard mirna/isomir naming (mirGFF3)
  • Module started by @oneillkza and completed by @FlorianThibord
• MultiVCFAnalyzer
  • Combining multiple VCF files into one coherent report and format for downstream analysis.
• Picard - new submodules for QualityByCycleMetrics, QualityScoreDistributionMetrics & QualityYieldMetrics
  • See #1116
• Rockhopper
  • RNA-seq tool for bacteria, includes bar plot showing where features map.
• Sickle
  • A windowed adaptive trimming tool for FASTQ files using quality
• Somalier
  • Relatedness checking and QC for BAM/CRAM/VCF for cancer, DNA, BS-Seq, exome, etc.
• VarScan2
  • Variant calling and somatic mutation/CNV detection for next-generation sequencing data

Module updates:

• BISCUIT
  • Major rewrite to work with new BISCUIT QC script (BISCUIT v0.3.16+)
    • This change breaks backwards-compatability with previous BISCUIT versions. If you are unable to upgrade BISCUIT, please use MultiQC v1.8.
  • Fixed error when missing data in log files (#1101)
• bcl2fastq
  • Samples with multiple library preps (i.e barcodes) will now be handled correctly (#1094)
• BUSCO
  • Updated log search pattern to match new format in v4 with auto-lineage detection option (#1163)
• Cutadapt
  • New bar plot showing the proportion of reads filtered out for different criteria (eg. too short, too many Ns) (#1198)
• DamageProfiler
  • Removes redundant typo in init name. This makes referring to the module's column consistent with other modules when customising general stats table.
• DeDup
  • Updates plots to make compatible with 0.12.6
  • Fixes reporting errors - barplot total represents mapped reads, not total reads in BAM file
  • New: Adds 'Post-DeDup Mapped Reads' column to general stats table.
• FastQC
  • Fixed tooltip text in Sequence Duplication Levels plot (#1092)
  • Handle edge-case where a FastQC report was for an empty file with 0 reads (#1129)
• FastQ Screen
  • Don't skip plotting % No Hits even if it's 0% (#1126)
  • Refactor parsing code. Avoids error with -0.00 %Unmapped (#1126)
  • New plot for Bisulfite Reads, if data is present
  • Categories in main plot are now sorted by the total read count and hidden if 0 across all samples
• fgbio
  • New: Plot error rate by read position from ErrorRateByReadPosition
  • GroupReadsByUmi plot can now be toggled to show relative percents (#1147)
• FLASh
  • Logs not reporting innie and outine uncombined pairs now plot combined pairs instead (#1173)
• GATK
  • Made parsing for VariantEval more tolerant, so that it will work with output from the tool when run in different modes (#1158)
• MTNucRatioCalculator
  • Fixed misleading value suffix in general stats table
• Picard MarkDuplicates
  • Major change - previously, if multiple libraries (read-groups) were found then only the first would be used and all others ignored. Now, values from all libraries are merged and PERCENT_DUPLICATION and ESTIMATED_LIBRARY_SIZE are recalculated. Libraries can be kept as separate samples with a new MultiQC configuration option - picard_config: markdups_merge_multiple_libraries: False
  • Major change - Updated MarkDuplicates bar plot to double the read-pair counts, so that the numbers stack correctly. (#1142)
• Picard HsMetrics
  • Updated large table to use columns specified in the MultiQC config. See docs. (#831)
• Picard WgsMetrics
  • Updated parsing code to recognise new java class string (#1114)
• QualiMap
  • Fixed QualiMap mean coverage calculation #1082, #1077
• RSeqC
  • Support added for output from geneBodyCoverage2.py script (#844)
  • Single sample view in the "Junction saturation" plot now works with the toolbox properly (rename, hide, highlight) (#1133)
• RNASeQC2
  • Updated to handle the parsing metric files from the newer rewrite of RNA-SeqQC.
• Samblaster
  • Improved parsing to handle variable whitespace (#1176)
• Samtools
  • Removes hardcoding of general stats column names. This allows column names to indicate when a module has been run twice (https://github.com/ewels/MultiQC/issues/1076).
  • Added an observed over expected read count plot for idxstats (#1118)
  • Added additional (by default hidden) column for flagstat that displays number total number of reads in a bam
• sortmerna
  • Fix the bug for the latest sortmerna version 4.2.0 (#1121)
• sexdeterrmine
  • Added a scatter plot of relative X- vs Y-coverage to the generated report.
• VerifyBAMID
  • Allow files with column header FREEMIX(alpha) (#1112)

Bug Fixes:

• Added a new test to check that modules work correctly with --ignore-samples. A lot of them didn't:
  • Mosdepth, conpair, Qualimap BamQC, RNA-SeQC, GATK BaseRecalibrator, SNPsplit, SeqyClean, Jellyfish, hap.py, HOMER, BBMap, DeepTools, HiCExplorer, pycoQC, interop
  • These modules have now all been fixed and --ignore-samples should work as you expect for whatever data you have.
• Removed use of shutil.copy to avoid problems with working on multiple filesystems (#1130)
• Made folder naming behaviour of multiqc_plots consistent with multiqc_data
  • Incremental numeric suffixes now added if folder already exists
  • Plots folder properly renamed if using -n/--filename
• Heatmap plotting function is now compatible with MultiQC toolbox hide and highlight (#1136)
• Plot config logswitch_active now works as advertised
• When running MultiQC modules several times, multiple data files are now created instead of overwriting one another (#1175)
• Fixed minor bug where tables could report negative numbers of columns in their header text
• Fixed bug where numeric custom content sample names could trigger a TypeError (#1091)
• Fixed custom content bug HTML data in a config file would trigger a ValueError (#1071)
• Replaced deprecated 'warn()' with 'warning()' of the logging module
• Custom content now supports section_extra config key to add custom HTML after description.
• Barplots with ymax set now ignore this when you click the Percentages tab.

A huge release, this one has been a long time coming. Due to @ewels being away on paternity leave for over six months it was very delayed and has been nearly a year in the making! During that time there has been 344 commits with 3,370 lines of code added and 1,194 deletions by 19 contributors. That's a lot of changes.

Highlights include:

• Finally removing the annoying YAML warning
• Six new modules, and many large updates to existing modules
• Code restructuring allowing MultiQC to be imported into Python environments and easier running on Windows
• Lots of tiny bug fixes all over the place.

Enjoy the update! And I promise I'll try not to make everyone wait so long for the next release...

Full changelog

New Modules:

• fgbio
  • Process family size count hist data from GroupReadsByUmi
• biobambam2
  • Added submodule for bamsormadup tool
  • Totally cheating - it uses Picard MarkDuplicates but with a custom search pattern and naming
• SeqyClean
  • Adds analysis for seqyclean files
• mtnucratio
  • Added little helper tool to compute mt to nuclear ratios for NGS data.
• mosdepth
  • fast BAM/CRAM depth calculation for WGS, exome, or targeted sequencing
• SexDetErrmine
  • Relative coverage and error rate of X and Y chromosomes

Module updates:

• bcl2fastq
  • Added handling of demultiplexing of more than 2 reads
  • Allow bcl2fastq to parse undetermined barcode information in situations when lane indexes do not start at 1
• BBMap
  • Support for scafstats output marked as not yet implemented in docs
• DeDup
  • Added handling clusterfactor and JSON logfiles
• damageprofiler
  • Added writing metrics to data output file.
• DeepTools
  • Fixed Python3 bug with int() conversion (#1057)
  • Handle varied TES boundary labels in plotProfile (#1011)
  • Fixed bug that prevented running on only plotProfile files when no other deepTools files found.
• fastp
  • Fix faulty column handling for the after filtering Q30 rate (#936)
• FastQC
  • When including a FastQC section multiple times in one report, the Per Base Sequence Content heatmaps now behave as you would expect.
  • Added heatmap showing FastQC status checks for every section report across all samples
  • Made sequence content individual plots work after samples have been renamed (#777)
  • Highlighting samples from status - respect chosen highlight colour in the toolbox (#742)
• FastQ Screen
  • When including a FastQ Screen section multiple times in one report, the plots now behave as you would expect.
• GATK
  • Refactored BaseRecalibrator code to be more consistent with MultiQC Python style
  • Handle zero count errors in BaseRecalibrator
• HiC Explorer
  • Fixed bug where module tries to parse QC_table.txt, a new log file in hicexplorer v2.2.
• HTSeq
  • Fixed bug where module would crash if a sample had zero reads (#1006)
• LongRanger
  • Added support for the LongRanger Align pipeline.
• miRTrace
  • Fixed bug where a sample in some plots was missed. (#932)
• Peddy
  • Fixed bug where sample name cleaning could lead to error. (#1024)
  • All plots (including Het Check and Sex Check) now hidden if no data
• Picard
  • Modified OxoGMetrics.py so that it will find files created with GATK CollectMultipleMetrics and ConvertSequencingArtifactToOxoG.
• QoRTs
  • Fixed bug where --dirs broke certain input files. (#821)
• Qualimap
  • Added in mean coverage computation for general statistics report
  • Creates now tables of collected data in multiqc_data
• RNA-SeQC
  • Updated broken URL link
• RSeQC
  • Fixed bug where Junction Saturation plot when clicking a single sample was mislabelling the lines.
  • When including a RSeQC section multiple times in one report, clicking Junction Saturation plot now behaves as you would expect.
  • Fixed bug where exported data in multiqc_rseqc_read_distribution.txt files had incorrect values for _kb fields (#1017)
• Samtools
  • Utilize in-built read_count_multiplier functionality to plot flagstat results more nicely
• SnpEff
  • Increased the default summary csv file-size limit from 1MB to 5MB.
• Stacks
  • Fixed bug where multi-population sum stats are parsed correctly (#906)
• TopHat
  • Fixed bug where TopHat would try to run with files from Bowtie2 or HiSAT2 and crash
• VCFTools
  • Fixed a bug where tstv_by_qual.py produced invalid json from infinity-values.
• snpEff
  • Added plot of effects

New MultiQC Features:

• Added some installation docs for windows
• Added some docs about using MultiQC in bioinformatics pipelines
• Rewrote Docker image
  • New base image czentye/matplotlib-minimal reduces image size from ~200MB to ~80MB
  • Proper installation method ensures latest version of the code
  • New entrypoint allows easier command-line usage
• Support opening MultiQC on websites with CSP script-src 'self' with some sha256 exceptions
  • Plot data is no longer intertwined with javascript code so hashes stay the same
• Made config.report_section_order work for module sub-sections as well as just modules.
• New config options exclude_modules and run_modules to complement -e and -m cli flags.
• Command line output is now coloured by default 🌈 (use --no-ansi to turn this off)
• Better launch comparability due to code refactoring by @KerstenBreuer and @ewels
  • Windows support for base multiqc command
  • Support for running as a python module: python -m multiqc .
  • Support for running within a script: import multiqc and multiqc.run('/path/to/files')
• Config option custom_plot_config now works for bargraph category configs as well (#1044)
• Config table_columns_visible can now be given a module namespace and it will hide all columns from that module (#541)

Bug Fixes:

• MultiQC now ignores all .md5 files
• Use SafeLoader for PyYaml load calls, avoiding recent warning messages.
• Hide multiqc_config_example.yaml in the test directory to stop people from using it without modification.
• Fixed matplotlib background colour issue (@epakarin - #886)
• Table rows that are empty due to hidden columns are now properly hidden on page load (#835)
• Sample name cleaning: All sample names are now truncated to their basename, without a path.
  • This includes for regex and replace (before was only the default truncate).
  • Only affects modules that take sample names from file contents, such as cutadapt.
  • See #897 for discussion.

An early Christmas present for MultiQC users! 🎅🎁🎄

Many thanks to everyone who has contributed to this release. Happy Christmas and a very happy new year!

New Modules:

• BISCUIT
  • BISuilfite-seq CUI Toolkit
  • Module written by @zwdzwd
• DamageProfiler
  • A tool to determine ancient DNA misincorporation rates.
  • Module written by @apeltzer
• FLASh
  • FLASH (Fast Length Adjustment of SHort reads)
  • Module written by @pooranis
• MinIONQC
  • QC of reads from ONT long-read sequencing
  • Module written by @ManavalanG
• phantompeakqualtools
  • A tool for informative enrichment and quality measures for ChIP-seq/DNase-seq/FAIRE-seq/MNase-seq data.
  • Module written by @chuan-wang
• Stacks
  • A software for analyzing restriction enzyme-based data (e.g. RAD-seq). Support for Stacks >= 2.1 only.
  • Module written by @remiolsen

Module updates:

• AdapterRemoval
  • Handle error when zero bases are trimmed. See #838.
• Bcl2fastq
  • New plot showing the top twenty of undetermined barcodes by lane.
  • Informations for R1/R2 are now separated in the General Statistics table.
  • SampleID is concatenate with SampleName because in Chromium experiments several sample have the same SampleName.
• deepTools
  • New PCA plots from the plotPCA function (written by @chuan-wang)
  • New fragment size distribution plots from bamPEFragmentSize --outRawFragmentLengths (written by @chuan-wang)
  • New correlation heatmaps from the plotCorrelation function (written by @chuan-wang)
  • New sequence distribution profiles around genes, from the plotProfile function (written by @chuan-wang)
  • Reordered sections
• Fastp
  • Fixed bug in parsing of empty histogram data. See #845.
• FastQC
  • Refactored Per Base Sequence Content plots to show original underlying data, instead of calculating it from the page contents. Now shows original FastQC base-ranges and fixes 100% GC bug in final few pixels. See #812.
  • When including a FastQC section multiple times in one report, the summary progress bars now behave as you would expect.
• FastQ Screen
  • Don't hide genomes in the simple plot, even if they have zero unique hits. See #829.
• InterOp
  • Fixed bug where read counts and base pair yields were not displaying in tables correctly.
  • Number formatting for these fields can now be customised in the same way as with other modules, as described in the docs
• Picard
  • InsertSizeMetrics: You can now configure to what degree the insert size plot should be smoothed.
  • CollectRnaSeqMetrics: Add warning about missing rRNA annotation.
  • CollectRnaSeqMetrics: Add chart for counts/percentage of reads mapped to the correct strand.
  • Now parses VariantCallingMetrics reports. (Similar to GATK module's VariantEval.)
• phantompeakqualtools
  • Properly clean sample names
• Trimmomatic
  • Updated Trimmomatic module documentation to be more helpful
  • New option to use filenames instead of relying on the command line used. See #864.

New MultiQC Features:

• Embed your custom images with a new Custom Content feature! Just add _mqc to the end of the filename for .png, .jpg or .jpeg files.
• Documentation for Custom Content reordered to make it a little more sane
• You can now add or override any config parameter for any MultiQC plot! See the documentation for more info.
• Allow table_columns_placement config to work with table IDs as well as column namespaces. See #841.
• Improved visual spacing between grouped bar plots

Bug Fixes:

• Custom content no longer clobbers col1_header table configs
• The option --file-list that refers to a text file with file paths to analyse will no longer ignore directory paths
• Sample name directory prefixes are now added after cleanup.
• If a module is run multiple times in one report, it's CSS and JS files will only be included once (default template)

Some of these updates are thanks to the efforts of people who attended the NASPM 2018 MultiQC hackathon session. Thanks to everyone who attended!

New Modules:

• fastp
  • An ultra-fast all-in-one FASTQ preprocessor (QC, adapters, trimming, filtering, splitting...)
  • Module started by @florianduclot and completed by @ewels
• hap.py
  • Hap.py is a set of programs based on htslib to benchmark variant calls against gold standard truth datasets
  • Module written by @tsnowlan
• Long Ranger
  • Works with data from the 10X Genomics Chromium. Performs sample demultiplexing, barcode processing, alignment, quality control, variant calling, phasing, and structural variant calling.
  • Module written by @remiolsen
• miRTrace
  • A quality control software for small RNA sequencing data.
  • Module written by @chuan-wang

Module updates:

• BCFtools
  • New plot showing SNP statistics versus quality of call from bcftools stats (@MaxUlysse and @Rotholandus)
• BBMap
  • Support added for BBDuk kmer-based adapter/contaminant filtering summary stats (@boulund
• FastQC
  • New read count plot, split into unique and duplicate reads if possible.
  • Help text added for all sections, mostly copied from the excellent FastQC help.
  • Sequence duplication plot rescaled
• FastQ Screen
  • Samples in large-sample-number plot are now sorted alphabetically (@hassanfa
• MACS2
  • Output is now more tolerant of missing data (no plot if no data)
• Peddy
  • Background samples now shown in ancestry PCA plot (@roryk)
  • New plot showing sex checks versus het ratios, supporting unknowns (@oyvinev)
• Picard
  • New submodule to handle ValidateSamFile reports (@cpavanrun)
  • WGSMetrics now add the mean and standard-deviation coverage to the general stats table (hidden) (@cpavanrun)
• Preseq
  • New config option to plot preseq plots with unique old coverage on the y axis instead of read count
  • Code refactoring by @vladsaveliev
• QUAST
  • Null values (-) in reports now handled properly. Bargraphs always shown despite varying thresholds. (@vladsaveliev)
• RNA-SeQC
  • Don't create the report section for Gene Body Coverage if no data is given
• Samtools
  • Fixed edge case bug where MultiQC could crash if a sample had zero count coverage with idxstats.
  • Adds % proper pairs to general stats table
• Skewer
  • Read length plot rescaled
• Tophat
  • Fixed bug where some samples could be given a blank sample name (@lparsons)
• VerifyBamID
  • Change column header help text for contamination to match percentage output (@chapmanb)

New MultiQC Features:

• New config option remove_sections to skip specific report sections from modules
• Add path_filters_exclude to exclude certain files when running modules multiple times. You could previously only include certain files.
• New exclude_* keys for file search patterns
  • Have a subset of patterns to exclude otherwise detected files with, by filename or contents
• Command line options all now use mid-word hyphens (not a mix of hyphens and underscores)
  • Old underscore terms still maintained for backwards compatibility
• Flag --view-tags now works without requiring an "analysis directory".
• Removed Python dependency for enum34 (@boulund)
• Columns can be added to General Stats table for custom content/module.
• New --ignore-symlinks flag which will ignore symlinked directories and files.
• New --no-megaqc-upload flag which disables automatically uploading data to MegaQC

Bug Fixes

• Fix path_filters for top_modules/module_order configuration only selecting if all globs match. It now filters searches that match any glob.
• Empty sample names from cleaning are now no longer allowed
• Stop prepend_dirs set in the config from getting clobbered by an unpassed CLI option (@tsnowlan)
• Modules running multiple times now have multiple sets of columns in the General Statistics table again, instead of overwriting one another.
• Prevent tables from clobbering sorted row orders.
• Fix linegraph and scatter plots data conversion (sporadically the incorrect ymax was used to drop data points) (@cpavanrun)
• Adjusted behavior of ceiling and floor axis limits
• Adjusted multiple file search patterns to make them more specific
  • Prevents the wrong module from accidentally slurping up output from a different tool. By @cpavanrun (see PR #727)
• Fixed broken report bar plots when -p/--export-plots was specified (see issue #801)

New Modules:

• DeDup - New module!
  • DeDup: Improved Duplicate Removal for merged/collapsed reads in ancient DNA analysis
  • Module written by @apeltzer,
• Clip&Merge - New module!
  • Clip&Merge: Adapter clipping and read merging for ancient DNA analysis
  • Module written by @apeltzer,

Module updates:

• bcl2fastq
  • Catch ZeroDivisionError exceptions when there are 0 reads (@aledj2)
  • Add parsing of TrimmedBases and new General Stats column for % bases trimmed (@matthdsm).
• BUSCO
  • Fixed configuration bug that made all sample names become 'short'
• Custom Content
  • Parsed tables now exported to multiqc_data files
• Cutadapt
  • Refactor parsing code to collect all length trimming plots
• FastQC
  • Fixed starting y-axis label for GC-content lineplot being incorrect.
• HiCExplorer
  • Updated to work with v2.0 release.
• Homer
  • Made parsing of tagInfo.txt file more resilient to variations in file format so that it works with new versions of Homer.
  • Kept order of chromosomes in coverage plot consistent.
• Peddy
  • Switch Sex error logic to Correct sex for better highlighting (@aledj2)
• Picard
  • Updated module and search patterns to recognise new output format from Picard version >= 2.16 and GATK output.
• Qualimap BamQC
  • Fixed bug where start of Genome Fraction could have a step if target is 100% covered.
• RNA-SeQC
  • Added rRNA alignment stats to summary table @Rolandde
• RSeqC
  • Fixed read distribution plot by adding category for other_intergenic (thanks to @moxgreen)
  • Fixed a dodgy plot title (Read GC content)
• Supernova
  • Added support for Supernova 2.0 reports. Fixed a TypeError bug when using txt reports only. Also a bug when parsing empty histogram files.

New MultiQC Features:

• Invalid choices for --module or --exclude now list the available modules alphabetically.
• Linting now checks for presence in config.module_order and tags.

Bug Fixes

• Excluding modules now works in combination with using module tags.
• Fixed edge-case bug where certain combinations of output_fn_name and data_dir_name could trigger a crash
• Conditional formatting - values are now longer double-labelled
• Made config option extra_series work in scatter plots the same way that it works for line plots
• Locked the matplotlib version to v2.1.0 and below
  • Due to two bugs that appeared in v2.2.0 - will remove this constraint when there's a new release that works again.

A slightly earlier-than-expected release due to a new problem with dependency packages that is breaking MultiQC installations since 2018-01-11.

New Modules:

• Sargasso
  • Parses output from Sargasso - a tool to separate mixed-species RNA-seq reads according to their species of origin
  • Module written by @hxin
• VerifyBAMID
  • Parses output from VerifyBAMID - a tool to detect contamination in BAM files.
  • Adds the CHIPMIX and FREEMIX columns to the general statistics table.
  • Module written by @aledj2

Module updates:

• MACS2
  • Updated to work with output from older versions of MACS2 by @avilella
• Peddy
  • Add het check plot to suggest potential contamination by @aledj2
• Picard
  • Picard HsMetrics HS_PENALTY plot now has correct axis labels
  • InsertSizeMetrics switches commas for points if it can't convert floats. Should help some european users.
• QoRTs
  • Added support for new style of output generated in the v1.3.0 release
• Qualimap
  • New Error rate column in General Statistics table, added by @Cashalow
    • Hidden by default - customise your MultiQC config to always show this column (see docs)
• QUAST
  • New option to customise the default display of contig count and length (eg. bp instead of Mbp).
  • See documentation. Written by @ewels and @Cashalow
• RSeQC
  • Removed normalisation in Junction Saturation plot. Now raw counts instead of % of total junctions.

New MultiQC Features:

• Conditional formatting / highlighting of cell contents in tables
  • If you want to make values that match a criteria stand out more, you can now write custom rules and formatting instructions for tables.
  • For instructions, see the documentation
• New --lint option which is strict about best-practices for writing new modules
  • Useful when writing new modules and code as it throws warnings
  • Currently only implemented for bar plots and a few other places. More linting coming soon...
• If MultiQC breaks and shows am error message, it now reports the filename of the last log it found
  • Hopefully this will help with debugging / finding dodgy input data

Bug Fixes

• Addressed new dependency error with conflicting package requirements
  • There was a conflict between the networkx, colormath and spectra releases.
  • I previously forced certain software versions to get around this, but spectra has now updated with the unfortunate effect of introducing a new dependency clash that halts installation.
• Fixed newly introduced bug where Custom Content MultiQC config file search patterns had been broken
• Updated pandoc command used in --pdf to work with new releases of Pandoc
• Made config table_columns_visible module name key matching case insensitive to make less frustrating

There are 34 merged pull-requests in this release - a fantastic example of how an open source community can develop a tool! Many thanks to everyone involved for their hard work.

Breaking changes - custom search patterns

Only for users with custom search patterns for the bowtie or star: you will
need to update your config files - the bowtie search key is now bowtie1,
star_genecounts is now star/genecounts.

For users with custom modules - search patterns must now conform to the search
pattern naming convention: modulename or modulename/anything (the search pattern
string beginning with the name of your module, anything you like after the first /).

New Modules:

• 10X Supernova
  • Parses statistics from the de-novo Supernova software.
  • Module written by @remiolsen
• BBMap
  • Plot metrics from a number of BBMap tools, a suite of DNA/RNA mapping tools and utilities
  • Module written by @boulund and @epruesse
• deepTools - new module!
  • Parse text output from bamPEFragmentSize, estimateReadFiltering, plotCoverage, plotEnrichment, and plotFingerprint
  • Module written by @dpryan79
• Homer Tag Directory - new submodule!
  • Module written by @rdali
• illumina InterOp
  • Module to parse metrics from illumina sequencing runs and demultiplexing, generated by the InterOp package
  • Module written by @matthdsm
• RSEM - new module!
  • Parse .cnt file comming from rsem-calculate-expression and plot read repartitions (Unalignable, Unique, Multi ...)
  • Module written by @noirot
• HiCExplorer
  • New module to parse the log files of hicBuildMatrix.
  • Module written by @joachimwolff

Module updates:

• AfterQC
  • Handle new output format where JSON summary key changed names.
• bcl2fastq
  • Clusters per sample plot now has tab where counts are categoried by lane.
• GATK
  • New submodule to handle Base Recalibrator stats, written by @winni2k
• HiSAT2
  • Fixed bug where plot title was incorrect if both SE and PE bargraphs were in one report
• Picard HsMetrics
  • Parsing code can now handle commas for decimal places
• Preseq
  • Updated odd file-search pattern that limited input files to 500kb
• QoRTs
  • Added new plots, new helptext and updated the module to produce a lot more output.
• Qualimap BamQC
  • Fixed edge-case bug where the refactored coverage plot code could raise an error from the range call.
• Documentation and link fixes for Slamdunk, GATK, bcl2fastq, Adapter Removal, FastQC and main docs
  • Many of these spotted and fixed by @juliangehring
• Went through all modules and standardised plot titles
  • All plots should now have a title with the format Module name: Plot name

New MultiQC Features:

• New MultiQC docker image
  • Ready to use docker image now available at https://hub.docker.com/r/ewels/multiqc/ (200 MB)
  • Uses automated builds - pull :latest to get the development version, future releases will have stable tags.
  • Written by @MaxUlysse
• New module_order config options allow modules to be run multiple times
  • Filters mean that a module can be run twice with different sets of files (eg. before and after trimming)
  • Custom module config parameters can be passed to module for each run
• File search refactored to only search for running modules
  • Makes search much faster when running with lots of files and limited modules
  • For example, if using -m star to only use the STAR module, all other file searches now skipped
• File search now warns if an unrecognised search type is given
• MultiQC now saves nearly all parsed data to a structured output file by default
  • See multiqc_data/multiqc_data.json
  • This can be turned off by setting config.data_dump_file: false
• Verbose logging when no log files found standardised. Less duplication in code and logs easier to read!
• New documentation section describing how to use MultiQC with Galaxy
• Using shared_key: 'read_counts' in table header configs now applies relevant defaults

Bug Fixes

• Installation problem caused by changes in upstream dependencies solved by stricter installation requirements
• Minor default_dev directory creation bug squashed
• Don't prepend the directory separator (|) to sample names with -d when there are no subdirs
• yPlotLines now works even if you don't set width