Question

Extract Features From Gbk Via Bash

0

Entering edit mode

10.6 years ago

Phil S. ▴ 700

Hi

is there any way to extract the locustag , join and complement information from a genbank via bash (one liner)? What gives me problems is the line break which may occur in the complement or join field of the genbank entry!

Thanks in advance.

Phil

ps. the lines I want to extract look like this:

/locus_tag="XXX_00002"


join(5485..5700,5784..6116,6230..6377,6434..6574,

6683..6819,7047..7314,7428..7938,8031..8307)


complement(join(34726..34766,34850..34866,34975..35287,

35392..35518,35604..35744,35831..35928,36051..36156))

bash genbank feature • 3.3k views

ADD COMMENT • link updated 10.6 years ago by xb ▴ 420 • written 10.6 years ago by Phil S. ▴ 700

2

Entering edit mode

Maybe, but it'd be a lot easier to use something like biopython or bioperl that can parse genbank files.

ADD REPLY • link 10.6 years ago by Devon Ryan 104k

0

Entering edit mode

historically this is why perl, python exists in the first place - so that we don't have to process text in bash

ADD REPLY • link 10.6 years ago by Istvan Albert 101k

0

Entering edit mode

Normally i solve stuff like this using boppython. The thing was that I wasn't working with my own laptop so i just needed to came up with a neat trick to solve it just the one time... Thanks for your suggestions anyways!

ADD REPLY • link 10.6 years ago by Phil S. ▴ 700

score 3 · Answer 1 · 2014-03-28

sed is one option.

The sample file,

cat sample.gbk

## output 
## ignored, same as in the question

To join the unwanted line breaks,

sed '/^$/d' sample.gbk | sed 'N;/,\n/s/"\? *\n//;P;D' 

## output
/locus_tag="XXX_00002"
join(5485..5700,5784..6116,6230..6377,6434..6574,6683..6819,7047..7314,7428..7938,8031..8307)
complement(join(34726..34766,34850..34866,34975..35287,35392..35518,35604..35744,35831..35928,36051..36156))

Then there are many ways to capture the contents. Below is one way to capture the "join" information,

sed '/^$/d' sample.gbk | sed 'N;/,\n/s/"\? *\n//;P;D' | sed -n 's/^join(\(.\+\))$/\1/p'

## output
5485..5700,5784..6116,6230..6377,6434..6574,6683..6819,7047..7314,7428..7938,8031..8307

Or all information in one line,

sed '/^$/d' sample.gbk | sed 'N;/,\n/s/"\? *\n//;P;D' | tr "\n" ";" | sed -r 's/^\/locus_tag="(.+)".*join\(([^()]+)\).*complement\(join\(([^()]+)\)\).*/\1\n\2\n\3\n/g'

## output
XXX_00002
5485..5700,5784..6116,6230..6377,6434..6574,6683..6819,7047..7314,7428..7938,8031..8307
34726..34766,34850..34866,34975..35287,35392..35518,35604..35744,35831..35928,36051..36156