v3.1.0

Added

• Support for Python 3.12.
• min_mask argument to GeneFinder to control the minimum lenght of masked regions on mask=True.

v3.0.1

Fixed

• Genes.write_scores and Genes.write_gff crashing on empty Genes (#44).

v3.0.0

Added

• MetagenomicBins collection to store a dense array of MetagenomicBin objects.
• metagenomic_bins keyword argument to GeneFinder allowing to control which models are used when running gene finding in meta mode (#24).
• metagenomic_bin attribute to Genes referencing the metagenomic model with which the genes were predicted, if in meta mode.
• Additional TrainingInfo properties (missing_motif_weight, coding_statistics).
• Setters for all remaining TrainingInfo properties.
• Proper TrainingInfo constructor with configuration option for all attributes.
• TrainingInfo.to_dict method to extract all parameters from a TrainingInfo.
• Genes.write_genbank method to write a GenBank record with all predicted genes from a sequence.
• include_stop flag to Gene.translate and Genes.write_translations to allow excluding the stop codon from the translated sequence.
• include_translation_table flag to Genes.write_gff to include the translation table to the GFF attributes of each gene.
• gbk output format to the Pyrodigal CLI.
• Sequence.unknown property exposing the number of unknown nucleotides in the sequence.
• Sequence.start_probability and Sequence.stop_probability to estimate the probability of encountering a start and a stop codon based on the GC%.

Fixed

• Genes.write_gff not properly reporting the number of bytes written.
• Merge several nogil sections in Sequence constructor.
• Several Cython functions missing a noexcept qualifier.

Changed

• BREAKING: Rename OrfFinder to GeneFinder for consistency.
• BREAKING: Use memoryview to expose all TrainingInfo attributes instead manually building lists or tuples.
• Reorganize memory management of the built-in metagenomic models.
• Make the internal Cython model public (pyrodigal.lib) to allow importing the underlying classes in other Cython projects.
• Use typing.Literal for allowed translation table values in pyrodigal.lib annotations
• Cache intermediate log-odds in Nodes._raw_coding_score to reduce calls to pow and log functions.
• Inline connection scoring functions to reduce function call overhead.
• Reorganize struct _node fields to reduce size in memory.
• Make GeneFinder.find_genes and GeneFinder.train reserve memory for the Nodes based on the GC% of the input sequence.
• Avoid storing temporary results in the generic implementation of ConnectionScorer.compute_skippable.
• Use Cython freelist for allocating Node, Gene, MetagenomicBin and Mask.
• Increase minimum allocation for Genes and Nodes to reduce early reallocations.

Removed

• BREAKING: metagenomic_bin attribute of TrainingInfo.

v3.0.0-alpha4

Added

• Sequence.unknown property exposing the number of unknown nucleotides in the sequence.
• Sequence.start_probability and Sequence.stop_probability to estimate the probability of encountering a start and a stop codon based on the GC%.

Changed

• Cache intermediate log-odds in Nodes._raw_coding_score to reduce calls to pow and log functions.
• Inline connection scoring functions to reduce function call overhead.
• Reorganize struct _node fields to reduce size in memory.
• Make GeneFinder.find_genes and GeneFinder.train reserve memory for the Nodes based on the GC% of the input sequence.
• Avoid storing temporary results in the generic implementation of ConnectionScorer.compute_skippable.

v3.0.0-alpha3

Fixed

• Merge several nogil sections in Sequence constructor.
• Several Cython functions missing a noexcept qualifier.

Changed

• Use Cython freelist for allocating Node, Gene, MetagenomicBin and Mask.
• Increase minimum allocation for Genes and Nodes to reduce early reallocations.

v3.0.0-alpha2

Added

• Genes.write_genbank method to write a GenBank record with all predicted genes from a sequence.
• include_stop flag to Gene.translate and Genes.write_translations to allow excluding the stop codon from the translated sequence.
• include_translation_table flag to Genes.write_gff to include the translation table to the GFF attributes of each gene.
• gbk output format to the Pyrodigal CLI.

Fixed

• Genes.write_gff not properly reporting the number of bytes written.

Changed

• Use typing.Literal for allowed translation table values in pyrodigal.lib annotations

v3.0.0-alpha1

Added

• MetagenomicBins collection to store a dense array of MetagenomicBin objects.
• metagenomic_bins keyword argument to GeneFinder allowing to control which models are used when running gene finding in meta mode (#24).
• metagenomic_bin attribute to Genes referencing the metagenomic model with which the genes were predicted, if in meta mode.
• Additional TrainingInfo properties (missing_motif_weight, coding_statistics).
• Setters for all remaining TrainingInfo properties.
• Proper TrainingInfo constructor with configuration option for all attributes.
• TrainingInfo.to_dict method to extract all parameters from a TrainingInfo.

Changed

• BREAKING: Rename OrfFinder to GeneFinder for consistency.
• Reorganize memory management of the built-in metagenomic models.
• Make the internal Cython model public (pyrodigal.lib) to allow importing the underlying classes in other Cython projects.
• BREAKING: Use memoryview to expose all TrainingInfo attributes instead manually building lists or tuples.

Removed

• BREAKING: metagenomic_bin attribute of TrainingInfo.

v2.3.0

Changed

• Bump Cython to v3.0.0.

v2.2.0

Changed

• Release GIL while masking sequence regions in Sequence.__init__.
• Use archspec instead of cpu_features for runtime feature detection.

Added

• CLI flag to run ORF detection in parallel when input contains several contigs.

Removed

• Support for Python 3.5.

v2.1.0

Changed

• Update Prodigal to v2.6.3+c1e2d36 to fix a bug with Shine-Dalgarno detection on reverse contig edge (hyattpd/Prodigal#100).

Added

• CLI flags to set the minimum gene size (#32, by @cjprybol).

Fixed

• ArchLinux User Repository package generation in CI.