How to extract the longest sequence from a fasta file

[clarissaboschi]

I have a fasta file with many sequences, an I would like to get just the longest one (does not matter the id). How can I do that?

I would like to have as output a fasta file with only the longest sequence.

thanks!

[Brian Bushnell]

You can make a read length histogram with BBMap's readlength.sh (or reformat.sh lhist), then run:

reformat.sh in=file.fa out=longest.fa minlength=X

[pmiguel]

Must. Not. Write. Perl. One. Liner.

--
Phillip

[kmcarr]

Originally Posted by clarissaboschi

I have a fasta file with many sequences, an I would like to get just the longest one (does not matter the id). How can I do that?

What about the case where there is more than one sequence which match the longest length? Save all? Only the first or last?

[Brian Bushnell]

Originally Posted by pmiguel

Must. Not. Write. Perl. One. Liner.

--
Phillip

Go right ahead

Originally Posted by kmcarr

What about the case where there is more than one sequence which match the longest length? Save all? Only the first or last?

These are interesting questions that make me wonder whether the original problem is worth solving. I'm guessing the OP wants the main chromosome from a perfect bacterial assembly, but perhaps the Clarissa should chime in on what she's trying to do.

[Biocomputronics]

This sort of task is nearly trivial in Python, it is well worth your while to learn it and the important packages such as BioPython which make jokes like seen in this thread all the more funny.

So here is a quick and dirty way. A little thought could make it more elegant and/or generalizable. This would just parse through your multifasta file and spit back the longest sequence.

from Bio import SeqIO

myList = []

for seq_record in SeqIO.parse("test.fa", "fasta"):
    myList.append([seq_record.id, str(seq_record.seq), len(seq_record)])

myList.sort(key=lambda x: x[2])

print("the longest sequence is:")
print(">", myList[-1][0], sep='')
print(myList[-1][1])

[clarissaboschi]

Thanks for the suggestions.
I have a list of protein ids, and I need to retrieve the respective nucleotide sequence. I am using eutils to retrieve them based on these ids, but for some ids, I have many sequences for one id. So for this specific case, I am selecting the longest sequence.

[Richard Finney]

/*
how to compile: gcc -Wall -O2 -o longestseq longestseq.c
example: ./longestseq example.fa

*/

#include <stdlib.h>
#include <string.h>
#include <stdio.h>

char s[100000]; // input buffer
int main(int argc, char *argv[])
{
FILE *fp;
int i;
size_t cur_spot;
size_t last_spot;
size_t best_spot;
size_t cursize;
size_t bestsofarsize;

if (argc != 2) { fprintf(stderr,"Error: need inputfilefasta. This program finds first longest fasta sequence(just one, no same length "ties"). Usage: ./longestseq example.fa\n"); exit(0); }
fp = fopen(argv[1],"r");
best_spot = cur_spot = last_spot = ftell(fp);
cursize = bestsofarsize = 0;
while (fgets(s,99999,fp))
{
for (i=0;s[i];i++) { if ((s[i]=='\r')||(s[i]=='\n')) {s[i] = (char)0;break;}} //handle dos or unix
if (s[0]=='>')
{
last_spot = cur_spot;
cursize = 0;
continue;
}
cursize += strlen(s);
if (cursize > bestsofarsize)
{
bestsofarsize = cursize;
best_spot = last_spot;
}
cur_spot = ftell(fp);
}

fseek(fp,best_spot,SEEK_SET);
fgets(s,99999,fp);
printf("%s",s);
while (fgets(s,99999,fp))
{
if (s[0] == '>') break;
printf("%s",s);
}
fclose(fp);
return 0;
}

[FONT="Courier New"][SIZE="1"]____