extract fasta sequences from multifasta file using partial or gene names

[SeqTroubles]

Hi All,

I am working PROKKA v1.12 files. I have a list of gene names such as

sacX
arcB
metB
sprT
adrB_2
fadD

and my fasta file is like so

>BOKHJPML_00001 hypothetical protein
ATGC
>BOKHJPML_00002 hypothetical protein
ATGC
>BOKHJPML_00003 Protease HtpX
ATGC
>BOKHJPML_00006 ATP-dependent Clp protease ATP-binding subunit ClpC
ATGC
BOKHJPML_00016 Inner membrane protein YfdC
ATGC

I want to extract the fasta sequences from the list. I have tried following previous suggestions using faidhttps://www.biostars.org/p/126204/x and biopyhttps://www.biostars.org/p/2822/thon
With no success. This faidx example is the closest I have come to success but I get a string of errors

warning: sacX not found in file
warning: arcB not found in file
warning: metB not found in file
warning: sprT not found in file
warning: adrB_2 not found in file
warning: fadD not found in file

Thanks in advance

[GenoMax]

One way would be to extract the full read headers from the sequence file using your ID's.

for i in `cat ./id_file `; do grep -i $i sequence.fa >> ID_in_sequence_file;done

Then use one of the methods you have found or faSomeRecords utility from Jim Kent to get the sequences extracted.

[Brian Bushnell]

You can also use BBMap's filterbyname.sh tool, particularly if you have a long list of names:

filterbyname.sh in=file.fa out=filtered.fa include=t names=names.txt substring

The "substring" flag allows partial matches.

[SeqTroubles]

You can also use BBMap's filterbyname.sh tool, particularly if you have a long list of names:

filterbyname.sh in=file.fa out=filtered.fa include=t names=names.txt substring

The "substring" flag allows partial matches.

Thanks so much Brian, this is very straight forward. Just a quick question about the tool. My gene name list contains some ambiguous names such as group_XXXX, as it is an output of roary. Would setting substring=t or substring=names cause it to partially match fasta headers from prokka output via the locus tag? If so is there a way to prevent this I have been using the following command:

filterbyname.sh in=seqs.ffn out=test.fasta include=t names=list.txt substring=name casesensitive=f

There are some seqs on my output which I feel should not be present. Although I do think the casesensitive flag is most likely the culprit?
Thanks.

[Brian Bushnell]

"substring=names" will consider a sequence to be a match if the sequence name contains any line in list.txt as a substring; and in this case, it's ignoring case. I suggest not ignoring case unless it's essential. Note that if you have any really short names in your file, like "A", it might match just about everything...