Generating genbanks with annotations for use with Clinker.
OK fine.
python generate_context_genbanks.py -nucdir [contigs directory] -protdir [protein gene cluster fasta directory] -pfamscandir [pfamscan output for the proteins you want] -outdir [output directory]
This script takes protein fastas containing gene clusters of interest and their corresponding nucleotide fasta files, combines them with multiple possible annotation file types, and constructs genbank files ready to use with Clinker (https://github.com/gamcil/clinker).
Because genbank files are a pain in the tuchus and figuring out the names of the subfields you can add annotations to is something you'd probably rather I do for you.
Importantly, the protein fastas in the protdir folder should only include the ORFs in the clusters you want to visualize with Clinker. This is important, so you should pay attention to this. Because it's important.
I'm assuming the nucleotide directory will be full of files called [genome_id].fna, the protein directory will be full of files called [genome_id].faa, pfamscan directory [genome_id].faa.out and kofamscan directory [genome_id].faa.out.parsed.good because of the custom pipeline I process KOFAMscan data with. More information follows, I suppose.
You can use multiple annotation file types simultaneously, since there are multiple fields clinker parses ORF information from in a genbank file.
-
kofamscan (https://github.com/takaram/kofam_scan)
-
GOOSOS (https://github.com/jwestrob/GOOSOS); specifically, the file
all_hits_evalues_dfthatGOOSOS.pygenerates after a successful run. -
Custom annotation files (see below example and the examples folder in this repository)
You talked about examples but I don't see any! I want to speak with your manager right now young man! Go get them. I'll wait.
Chill Karen. I need to make a gene cluster fasta and generate associated annotations for a public-domain genome that's not part of one of my ongoing projects, so give me a hot second OK?
I keep trying to use kofamscan but your script gets mad at me or messes up in some miscellaneous way.
I'm not surprised; in the Banfield lab we've developed a processing pipeline that takes the output of KOFAMscan and selects only the above-threshold hits. If you're trying to use this script, want to include KOFAM info, and aren't in the Banfield lab, hit me up.
Yes indeed. Please make this a tab separated, two column file with the first column being the protein IDs and the second the custom annotation you wish to display. See this example:
example_genome_id|example_orf_id_1 example_annotation_1
example_genome_id|example_orf_id_2 example_annotation_2
You might notice that I have the genome ID in the protein ID, separated with a | character. As of writing this (the first day of writing this script), I haven't explored a good number of edge cases, so if you don't have those genome IDs appended, there may be some key error issues. I'm going to make sure the script is robust to this in the future, but for now, caution when making custom annotations if you do not have the genome ID included.
Yeah I haven't been working on this one for very long, and I'm sure there are a bunch of ways to break it since I'm dealing with numerous different file formats. If you experience problems send me a slack (if you're in the Banfield lab) or raise an issue here and I'll fix it PDQ.