pyrodigal

Fixed

• GC% computation and RBS scoring for reverse strand nodes close to the contig edge (#27).

Fixed

• OrfFinder(mask=True) ignoring the minimum mask size when masking regions (#26).

Changed

• Use cibuildhweel for building wheel distributions.

Added

• Wheels for MacOS Aarch64 platforms.

Fixed

• Syntax issue in Cython files failing build on Bioconda runner.

Fixed

• Syntax issue in Cython files failing build on some environments.

Added

• MMX implementation of the SIMD prefilter.
• Proper GFF headers and metadata section to GFF output.
• Sequence.gc_frame_plot method to compute the max GC frame profile from Python.
• metagenomic_bin property to TrainingInfo to support recovering the object corresponding to a pre-trained model.
• meta attribute to Genes to store whether genes were predicted in single or in meta mode.
• pyrodigal.PRODIGAL_VERSION constant storing the wrapped Prodigal version.
• pyrodigal.MIN_SINGLE_GENOME and pyrodigal.IDEAL_SINGLE_GENOME constants storing the minimum and recommended sequence sizes for training.

Changed

• Make all write methods of Genes objects require a sequence_id argument instead of using the internal sequence number.
• Rewrite SIMD prefilter using a generic template with C macros.
• Make Mask record coordinates in start-inclusive end-exclusive mode to follow Python conventions.
• Make connection scoring tests only score some randomly selected node pairs for faster runs.
• Rewrite tests to use importlib.resources for managing test data.

Removed

• from_bytes and from_string constructors of Sequence objects.

Fixed

• Duplicate extraction of start codons located on contig edges inside Nodes._extract (#21).
• Pickling and unpickling of TrainingInfo objects corresponding to pre-trained models.
• Implementation of calc_most_gc_frame being inconsistent with the Prodigal implementation.
• Implementation of the maximum search in score_connection_forward_start not following the (weird?) behaviour from Prodigal (#21).
• Gene identifier being used instead of the sequence identifier in the GFF output (#18).
• Out of bound access to sequence data in Sequence._shine_dalgarno_mm and Sequence._shine_dalgarno_exact.

Changed

• Make Mask record coordinates in start-inclusive end-exclusive mode to follow Python conventions.

Removed

• from_bytes and from_string constructors of Sequence objects.

Added

• Sequence.gc_frame_plot method to compute the max GC frame profile from Python.

Changed

• Rewrite tests to use importlib.resources for managing test data.
• Make connection scoring tests only score some randomly selected node pairs for faster runs.

Fixed

• Duplicate extraction of start codons located on contig edges inside Nodes._extract.

Added

• metagenomic_bin property to TrainingInfo to support recovering the object corresponding to a pre-trained model.
• meta attribute to Genes to store whether genes were predicted in single or in meta mode.

Fixed

• Pickling and unpickling of TrainingInfo objects corresponding to pre-trained models.
• Implementation of calc_most_gc_frame being inconsistent with the Prodigal implementation.
• Implementation of the maximum search in score_connection_forward_start not following the (weird?) behaviour from Prodigal.

Added

• MMX implementation of the SIMD prefilter.
• Propert GFF headers and metadata section to GFF output.

Fixed

• Out of bound access to sequence data in Sequence._shine_dalgarno_mm and Sequence._shine_dalgarno_exact.
• Gene identifier being used instead of the sequence identifier in the GFF output (#18).

Changed

• Rewrite SIMD prefilter using a generic template with C macros.
• Make all write methods of Genes objects require a sequence_id argument instead of using the internal sequence number.

Changed

• Use the vbicq Arm intrinsic in the NEON implementation to combine vandq and vmvnq.

Fixed

• Prevent direct instantiation of Node and Gene objects from Python code.
• Configuration of platform-specific NEON flags in setup.py not being applied to the linker.

Fixed

• Some cpu_features source files not being included in source distribution.

Changed

• OrfFinder.train can now be given more than one sequence argument to train on contigs from an unclosed genome.
• Updated cpu_features to v0.7.0 and added hardware detection of NEON features on Linux Aarch64 platforms.

Fixed

• Detection of Arm64 platform in setup.py (#16).

Changed

• pyrodigal.cli now concatenates training sequences the same way as Prodigal does.

Stable version, to be published in the Journal of Open-Source Software.

Added

• pickle protocol implementation for Nodes, TrainingInfo, OrfFinder, Sequence, Masks and Genes objects.
• Buffer protocol implementation for Sequence, allowing access to raw digits.
• __eq__ and __repr__ magic methods to Mask objects.

Changed

• Optimized code used for region masking to avoid searching for the same mask repeatedly.
• TRANSLATION_TABLES and METAGENOMIC_BINS are now exposed as constants in the top pyrodigal module.
• Refactored connection scoring into different functions based on the type (start/stop) and strand (direct/reverse) of the node being scored.
• Changed the growth factor for dynamic arrays to be the same as the one used in CPython list buffers.

Added

• Gene.score property to get the gene score as reported in the score data string.

Fixed

• OrfFinder.find_genes not producing consistent results across runs in meta mode (#13).
• OrfFinder.find_genes returning Nodes with incomplete score information.

Changed

• Improve performance of mer_ndx and score_connection using dedicated implementations with better branch prediction.
• Mark arguments as const in C code where possible.

Fixed

• Signatures of Cython classes not displaying properly because of the embedsignature directive.
• _sequence.h functions not being inlined as expected.

Changed

• Rewrite internal Sequence code using inlined functions to increase performance when the strand is known.

Fixed

• Nodes.copy potentially failing on empty collections after trying to allocate 0 bytes.
• TestGenes.test_write_scores failing on some machines because of float rounding issues.
• Gene.translate ignoring the unknown_residue argument value and always using "X".
• Memory leak in Pyrodigal.train cause by memory not being freed after building the GC frame plot.

Added

• Support for setting a custom minimum gene length in pyrodigal.OrfFinder.
• Genes.write_scores method to write the node scores to a file.
• Gene.__repr__ and Node.__repr__ methods to display some useful attributes.
• Sequence.__str__ method to get back a nucleotide string from a Sequence object.

Changed

• Use a more compact data structure to store Gene data.

Fixed

• Nodes._calc_orf_gc reading nucleotides after the sequence end when computing GC content for edge nodes.

Removed

• pyrodigal.Pyrodigal class (use pyrodigal.OrfFinder instead).
• pyrodigal.Predictions class (functionality merged into pyrodigal.Genes).

Added

• load and dump methods to TrainingInfo for storing and loading a raw training info structure.
• Support for creating an OrfFinder pre-configured with a training info.
• -t and -n flags to the CLI.