Future updates/fixes to consider!

[raufs]

This issue/ticket is to track ideas for downstream improvement of the suite:

• Extract more information from HyPhy reports to include in the consolidated spreadsheets. #97
• Add options for skani sketching/searching, in particular for marker k-mer compression factor. #75
• - Switch to pyhmmer in annotation step for HMM based databases to more effectively use available threads. #18
• By default highly similar genes in a gene-cluster which do not exhibit similarity to genes from other gene-clusters are regarded as paralogs and thus partitioned into separate ortholog groups. Consider merging such cases into a common ortholog group if high sequence similarity exists.
• Incorporate an option for automatic visualization options for dereplicated gene cluster instances in zol (e.g. using pygenomeviz or clinker) #76
• Correct FASTA output for near-core representative proteins for fai parameter suggestions in zol and handling of cases when 0 or only 1 core proteins are found (will be addressed in v1.5.6). #88
• Add flag to avoid replacing start proteins with M if they are instead L or V based on direct translation (implemented with bacteria in mind) if eukaryotic analysis is being performed (will be addressed in v1.5.6). #89
• Correct generateSyntenicVisual.py usage with new setup which generates R scripts (will be addressed in v1.5.6). #90
• Update cgcg to also work with results from "domain mode".
• Add options to control inflation parameter for MCL and to perform simple single-linkage clustering for ortholog group delineation. #91
• Improve fai help function around providing multiple query GenBank(s) - make clear these should be highly related instances (e.g. those belonging to the same GCF) and add documentation on wiki for describing what ends up happening behind the algorithm (e.g. queries are not treated independently but rather merged).

Please feel free to add suggestions or make pull requests if you want to implement them yourselves!

[raufs]

• Improve handling / preparation of rep_to_nonrep_prots dictionary to reduce memory.

• remove pympler import and dependency for bioconda.

• consider configuring numba version on macos to avoid deprecated warning: https://stackoverflow.com/questions/73278739/omp-set-nested-routine-deprecated-please-use-omp-set-max-active-levels-instead

• update abon parsing of antiSMASH GenBank files to make parsing more robust and work on anitSMASH >8 results.

• update fai's logic for preliminary screening to also account for key proteins needed

• consider using libcurl instead of axel for downloading.

[raufs]

Major next updates planned:

• Consider doing layered searching whereby fai with certain queries can be used to create a smaller, more limited, prepTG database only containing genomes which contain those queries. Then another query / fai run can be performed after this on the limited prepTG database - which should run faster.

• Have an option for on the fly GenBank creation for metagenomic datasets after gene cluster detection (in fai) to avoid creating large-space costly GenBank files for full (meta-)genomes in advance (in prepTG) to extract gene clusters from. Instead, with certain options contained, prepTG will harbor more light-weight annotation formats, like Prokka-like GFFs (TSV followed by FASTA) or GenBank files without translations.

• Incorporate peptides.py into zol reports

• Switch from CD-HIT to diamond linclust in zol both for re-inflation and low-mem option for homology collapsing.

• Implement calculation of McDonald-Kreitman test when comparative mode requested in zol.

• Consider adding the following to mute warnings about locus line length produced by Biopython during GenBank creation steps when scaffolds have long names. Users should be able to ignore this and long LOCUS lines appear to be properly parsed by Biopython.

import warnings
warnings.filterwarnings("ignore", message="Increasing length of locus line")

• Switch from Biopython to gb-io for GenBank parsing.

• Switch or provide as an option FAMSA instead of MUSCLE for multiple sequence alignment.