- Produce gff files in addition to gbk for onegene, standardize, and postprocessing
- Write a log file to the output folder (#73)
- Fixed verbose and quiet run modes.
- The criteria for accepting coordinate corrections now also requires valid start codons. (#71)
- Fixed off by 1 error when reporting the location of translational exceptions (selenocysteine proteins). (#72)
- Lifted memory restrictions on cd-hit (#74)
- Fixed crash that occurs when only one gene component of a fusion doesn't have a product field. (#75)
- hybran now depends on
networkxandintervaltree
- New subcommand:
hybran compare: compare two annotations of the same genome
- Rewritten comparison engine based on intervaltrees. This fixes some edge cases, such as pairs of in-frame overlapping genes with an unrelated gene between them, as well as unannotated regions at contig edges. The code is modular, available to run in the above-mentioned new command, and used in annomerge.
- Enhancements to
hybran standardize:- New option
-r/--ref-names-only. - Simplified usage: If passing in a hybran output directory, no other arguments will be necessary.
- New option
- Annotations now include pseudoscan evidence codes as note fields. These were previously only found in the pseudoscan report files.
- Annomerge no longer produces redundant CDS entries (#69)
- Synonymous SNPs in start codons do not automatically invalidate reference-correspondence.
- pseudoscan: Fixed typo potentially affecting calls where the reference is pseudo (#68).
- fusionfisher:
- Identify another signature of hybrid fusion genes (previously identified as misannotations)
- Update FeatureProperties for fusion genes based on their components. Fusions no longer categorically pseudo.
- Eliminated more false positive fusion calls by accounting for another misannotation signature.
- Coordinate corrections no longer collapse compound locations ("joins") into single intervals (#70).
- Include handling of
transl_exceptqualifiers for amino acids introduced through translational recoding. - Restored compatibility with biopython >=1.82 and applied more preventitive maintenance in response to deprecation warnings.
- Translations, in particular for RATT-transferred annotations, fixed to account for alternative methionine start codons.
- Recognize .gbff as a genbank format file extension.
- Write out all Prokka postprocessed contigs' annotations instead of just the last one.
- Fixed issue with
onegenewhere unnamed reference genes did not get unified to use the locus tag of the primary reference as a gene name, instead unnecessarily assigning a generic name.
- New subcommands:
hybran standardize: remove generic gene names in the final annotations.hybran onegene: unify names of highly conserved gene copies
- New option
-s/--organismto set organism name in the final genbank header (#58). - Intercept malformed features in RATT output to allow continuity of the pipeline (#64).
- The
--dedupe-referencesoption is now deprecated, being made a core part of the pipeline withonegenesince the generic names that it introduces to collapse reference paralogs can be undone byhybran standardizeif desired afterwards. - Reference unification now uses a different generic prefix than that used for unnamed genes to allow differentiating between them.
hybran standardizehandles these as well. - Full support for multiple reference annotations.
- Postprocessed versions of RATT and ab initio annotations, with associated reports, are now saved.
- Substantial speedups due to parallelization of all postprocessing logic.
- Logging and reporting:
- Comprehensive reporting of pseudoscan results.
- Reorganized the invalid and rejected features reports.
- Removed spaces from column names of novelty report.
- reference unification (formerly activated by the
--dedupe-referencesoption):- fixed for multiple references (#55)
- enabled checkpointing on this step in the full pipeline.
- pseudoscan: now identifies delayed stop codons when start coordinate correction changes the reading frame without introducing an internal stop.
- coord_check:
- Fixed issue in coord_check's reporting status when not attempting correction.
- Coordinate correction sometimes resulted in genes with no stop codons. coord_check now checks for this and extends the ORF to the next in-frame stop codon to make a proper correction. These address the root cause of the problem for which the temporary measure from v1.6.1 was taken. That temporary measure has been removed.
- Improved detection of gene fusions when one of the components is derived from a reference pseudogene.
- fissionfuser: fixed issue where apparent complementary fragments are combined despite one of the copies being non-pseudo (#66). Although the issue severity was mitigated in that the combined annotation would be rejected in favor of RATT's, the second gene would have been lost.
- annomerge: reject the
sourcefeature from RATT. Some final annotations contained two of them: one from RATT and one from Prokka. - Fixed issues that occur when sequence IDs contain "|" character (#62).
- Fixed handling of situations where either RATT or Prokka find no annotations.
- Fixed problem with redundant fusion gene name components when detected using both RATT and Prokka.
- Improved detection of gene fusions due to adjustments of alignment internal gap extension penalty and refined delayed-stop calling criteria.
- Changed default generic ORF prefix to "HYBRA" for greater clarity.
- set default output directory to current directory.
- Now ignoring warnings when generating translations for pseudogenes that aren't multiples of three.
- Enabled setting any of RATT's configured transfer types and fixed names for *.global parameter sets.
- Fixed scenario where hybran crashes when coordinate correction matches to an adjacent locus and attempts correction (#61).
- Added a temporary measure to inclusion criteria to penalize CDSs lacking stop codons (inadequately postprocessed) (#60).
- Massive streamlining of the pipeline. Reworked components into new subsystems:
pseudoscan: identification of anomalous copies of reference genes using new criteria independent of alignment coverage. (#50, #56, and #59)fissionfuser(formerlyprocess_split_genes()): improved detection and combining of gene fragments (that ab initio annotations tend to produce) into a single record.fusionfisher: detection of gene fusion events and putative misannotations.thunderdome: more aggressive conflict resolution between RATT and ab initio annotations.
- Output GFF files no longer include the genome sequence.
- Fixed handling of conflicting annotations that are differently named (#57).
- Reimplemented coordinate correction and applied to ab initio ORFs as part of
pseudoscan. This resolves many instances of falsepseudoCDSs ab initio that were due simply to incorrect start coordinate predictions spuriously shortening the genes. - Fixed handling of compound intervals in reference annotations (#46, #47)
- Resolved issues involving reference annotations with multiple contigs/chromosomes (#48)
- Fixed issue with some gene name assignments being dropped later in the pipeline due to some obsolete code (#43).
- More comprehensive tracking of RATT/ab initio overlaps and conflicts (#49).
- Checking in-frame overlaps with pseudo ORFs containing internal stop codons
- Revamped postprocessing of RATT-introduced compound intervals (#44, #45)
- Updated inclusion criteria for special handling of pseudo ORFs (#42)
- Made a consistent non-CDS policy for RATT: Take everything except rRNA and tRNA (#22)
- Clarified some rejection reasons for RATT/ab initio features.
- Fixed representation of blast results for CDSs when there aren't any hits
- Fixed issue with RATT handoff if sample/contig names contain
.or|. - Fixed issue with logging merged genes.
- Fixed newly introduced issue with rejecting RATT annotations.
- Removed some unused code.
- Prevented overlapping RATT-transferred annotations from automatically being handled as conflicts, leading one of the two to be discarded (#39)
- Corrected distinguishing of ab initio vs reference-transferred annotation in final conflict resolution step (#38)
- Added logging of some missed cases of annotation rejections (for {ratt,prokka}_unused.tsv)
- Fixed the reference gene <=> locus_tag mapping dictionaries used in annomerge (#41)
- Made sure to track and process all ab initio annotations that overlap RATT-transferred CDSs (#40).
- Fixed handling of multi-fasta inputs (#35)
- Streamlined Prokka workflow (#33)
- Parallelized BLASTing to reference genes in annomerge.
- Set sequence names in the output annotation files.
- Added exit status checks so pipeline fails as early as possible when things go wrong.
- Switched default evalue to Prokka's current setting of 1e-9 (was 1e-6)
- Slightly streamlined the RATT / Prokka comparison workflow
- Added more unit tests
- Genes split into multiple adjacent fragments used to have a single /gene record but multiple CDS records with the same locus tag. For INSDC compliance, they now only have a single CDS record as well.
- Removed /translation fields for /pseudo CDSs.
- Generalized for any prokaryote.
- Genetic code and taxonomy ID detected from reference annotation. RATT configuration is now automatically generated based on the detected genetic code, so a configuration file is no longer bundled.
- Now using "ORF" prefix for generic genes rather than "MTB" by default.
Option
--orf-prefixadded for customizability.
- Removed checking for dnaA as the first gene at the first base position.
- Made eggNOG-mapper step optional.
- Gene fragments are now identified using the corresponding reference gene names, but are distinguished with a /pseudo tag.
- RATT and (some) Prokka options are now under user control.
- Account for translationless CDSs that are labeled with the 'pseudo' qualifier instead of 'pseudogene'
- Allow input fasta files with alternative standard extensions.
- Fixed handling of reference annotations that may not have /gene qualifiers for all annotations.
- Fixed handling of input genome when it's the same as the reference.
- Set proper field from which to draw eggNOG-mapper annotations.
- Uniform locus tags are now assigned for every sample.
- Better identification of reference and unnamed genes in processing of clusters.
- Hybran version now recorded in the genbank annotation header.
- Updated the Prokka reference proteome generation format to enable Prokka to set gene names and product fields rather than leaving it to the final clustering step.
- Fixed installation location of resource file.
- Added
--dedupe-referencesoption to assign a single generic gene name to duplicate genes in the provided reference annotations. - Sequence identity and alignment coverage thresholds are no longer applied to RATT-transferred annotations by default. (#28)
The original behavior can be restored by passing the new
--filter-rattoption. - reference annotations can now be passed as individual file names or file of file names, in addition to a directory name (#21)
- eggnog-mapper step no longer gets skipped (#23)
- alignment query coverage threshold is now applied directly in Prokka (#24)
- Dropped criterion of excluding hypothetical genes from Prokka-no-reference (#30)
- Fixed calculation of query and reference alignment coverage (#27).
- Corrected selection of top blastp hits for the one-to-one and one-to-many searches. When there were multiple hits in these cases, only the last one output by BLAST was being retained, which actually corresponds to the worst hit (by e-value). We now retain only the first hit.
- Add thorough logging of gene annotations merged, rejected, and newly-named. (#19)
- Better tolerance of directory name inputs (#18)
- New option
-c/--coverage-thresholdfor tuning gene matching. The sequence identity threshold is now taken through-i/--identity-threshold. - Incorporate a RATT configuration file for using codons from translation table 11 (#17)
- use provided identity/coverage thresholds for all instances of BLAST, CD-HIT (#10)
- ensure that the final gff output file gets updated (#15)
- (#14)
- Fixed issue with writing of merged_genes.gbk that sometimes caused hybran to crash during a run.
- Renamed to Hybran
- Migrated to Python 3
- Removed limitation of 30 references
- Allowed sequence identity threshold to be user-defined
- Proper handling of temporary files
- Fixed issue preventing clustering step from running
Initial release of AnnoTUB