Trying to use python regular expressions to filter fasta file sequence headers

Question

[Ash]

Hi,

Apologies if I do not follow the correct question formatting, this is my first time posting. My question is regarding the use of python regular expressions. I have a fasta file of sequences following the format:

>NODE_143195_length_100_cov_16076.000000
TTGTGTTGGTTGTTGTGTTGCCTGTCTTGGTGGCGGTTGTGTTGGCTGCTTTCGTGTCAG
TCTCTTCACCGATGTTATGTTGCTCTGTTGTGGCTCCGGC
>NODE_143196_length_100_cov_15891.000000
CTTGTGTTGGTTGTTGTGTTGCCTGTCTTGGTGGCGGTTGTGTTGGCTGCTTTCGTGTCA
GTCTCTTCACCGATGTTATGTTGCTCTGTTGTGGCTCCGG
>NODE_143197_length_100_cov_15696.000000
GCTTGTGTTGGTTGTTGTGTTGCCTGTCTTGGTGGCGGTTGTGTTGGCTGCTTTCGTGTC
AGTCTCTTCACCGATGTTATGTTGCTCTGTTGTGGCTCCG

I am trying to filter by both length and coverage. I want to filter sequences less than 5000bp and less than 100 coverage. I have been trying different variations of the following line:

^.+cov=([5-9][0-9][0-9]|([1-9]\d{4}\d*)\..+$

But I cannot seem to make it work. If anyone can help me, would be greatly appreciated. Thanks

Answer 1

[cpad0112]

Are the length and coverage values taken from header? If so, keep it simple:

from Bio import SeqIO
import sys

for i in SeqIO.parse("file.fa","fasta"):
     isplit=i.id.rsplit("_")
     if (int(isplit[3])>20 and int(float(isplit[5]))>100) :
         print(">"+i.id,"\n",i.seq)

input and output:

$ cat file.fa 
>NODE_143195_length_10_cov_16000.000000
TTGTGTTGGTTGTTGTGTTGCCTGTCTTGGTGGCGGTTGTGTTGGCTGCTTTCGTGTCAGTCTCTTCACCGATGTTATGTTGCTCTGTTGTGGCTCCGGC
>NODE_143196_length_100_cov_15891.000000
CTTGTGTTGGTTGTTGTGTTGCCTGTCTTGGTGGCGGTTGTGTTGGCTGCTTTCGTGTCAGTCTCTTCACCGATGTTATGTTGCTCTGTTGTGGCTCCGG
>NODE_143197_length_100_cov_15.000000
GCTTGTGTTGGTTGTTGTGTTGCCTGTCTTGGTGGCGGTTGTGTTGGCTGCTTTCGTGTCAGTCTCTTCACCGATGTTATGTTGCTCTGTTGTGGCTCCG

$ python test.py file.fa
>NODE_143196_length_100_cov_15891.000000 
 CTTGTGTTGGTTGTTGTGTTGCCTGTCTTGGTGGCGGTTGTGTTGGCTGCTTTCGTGTCAGTCTCTTCACCGATGTTATGTTGCTCTGTTGTGGCTCCGG

You can use seqio.write instead of print as the object after parsing is still seqrecord.