remove part of a multifasta header with sed or other

Question

[tremblayemilie9]

Hi!

I have something like this:

>barcodelabel=#ITS2_A_B1_VG6RM_00076_01732;size=12594;
TCGTTCTCGGACTTTGGGTACAAGGGGCAGGGCTGGCTGCTTCCGGCAGGCGGCCCCGCCGGCGGCGGGGGCCGCCAGTC
GCCGAGTCCTGGCCGCGGTTGCAAAGGGTGGGGTGGCGCCCGGGGGCGTGACCCATTAATGATCCTTCCGCAGGTTCACC
TACGGAAACCTTGTTACGACTTTTACTTCCTCTAAATGACCAAG
>barcodelabel=#ITS2_A_B2_VG6RM_00466_00157;size=9208;
TTAAGTTTTTTCAGACGCTGATTGCAACTGCAAATGGTTTAAATTGTCCAATCGGCGGGCGGACCCGCCGAGGAAACGTA
AGGTACTTAAAAGACATGGGTAAGAGATAGCAGGCAAAGCCTACAACTCTAGGTAATGATCCTTCCGCAGGTTCACCTAC
GGAAACCTTGTTACGACTTTTACTTCCTCTAAATGACCAAG
>barcodelabel=#ITS2_A_B1_VG6RM_00284_01321;size=8857;
TTAATTTGTTACTGACGCTGATTGCAATTACAAAAGGTTTATGTTTGTCCTAGTGGTGGGCGAACCCACCAAGGAAACAA
GAAGTACGCAAAAGACAAGGGTGAATAATTCAGCAAGGCTGTAACCCCGAGAGGTTCCAGCCCGCCTTCATATTTGTGTA
ATGATCCCTCCGCAGGTTCACCTACGGAGACCTTGTTACGACTTTTACTTCCTCTAAATGACCAAG

and I would like to take off the read id part (VG6RM_00284_01321) with sed.

I do not know how do do so since the part before varies (B1, B2, etc).

Thank you

Answer 1

[iraun]

If you have only B one time (I mean B1, B2, B3 ... but only once, not B1_B2 in the same read), you can try this to extract the read ID:

sed -r 's/(^>.*B[0-9]\_)[^;]*(;.*)/\1\2/' file >out_file

[Ram]

That seems a bit off the mark to me. Shouldn't it be

sed -r 's/(^>.*B[0-9]\_)[^;]*(;.*)/\1\2/' file >out_file

as the aim is to retain everything except the part between B[0-9]_ and ; ?

[iraun]

Oh, I understood the opposite, maybe you're right RamRS, but I'm not sure yet.

[iraun]

But it is a bit strange to keep everything except the read ID, isn't it?

[Ram]

Maybe, but that's what OP wants to do. No idea why.

[Ram]

Also, you might wanna edit your answer so OP can accept it.

[iraun]

RamRS, you won the battle ;)

[iraun]

It is the correct way to do? I mean, the correct answer is yours, not mine, so I guess that you should post your comment as an answer?

[Ram]

It's OK. Virtual points are a petty thing to compete for. Plus, you misunderstood the question, so it's not like you did not know how to get there :-)

Also, OP has accepted your answer. High time for you to edit the content!

[iraun]

hahaha, OK, thank you :)

[Ram]

Also, to circumvent your "if there is only one B1/B2", you can use a minimal/non-greedy match expression. Over here, a greedy match of the \1 expression serves the purpose, but let's say you wanna match the shortest a*b in "aabb", if you use the expression a*?b, the match is aab (as opposed to the aabb match for the usual a*b expression - the ? makes all the difference!)

[tremblayemilie9]

Hi, I did your command line (sed -r 's/(>*).*_B[0-50]_([^;]+).*/\1\2/'g test.fasta > out_test_.fasta) but it actually does the opposite of what I want:

>VG6RM_00466_00157
TTAAGTTTTTTCAGACGCTGATTGCAACTGCAAATGGTTTAAATTGTCCAATCGGCGGGCGGACCCGCCGAGGAAACGTA
AGGTACTTAAAAGACATGGGTAAGAGATAGCAGGCAAAGCCTACAACTCTAGGTAATGATCCTTCCGCAGGTTCACCTAC
GGAAACCTTGTTACGACTTTTACTTCCTCTAAATGACCAAG
>VG6RM_00284_01321
TTAATTTGTTACTGACGCTGATTGCAATTACAAAAGGTTTATGTTTGTCCTAGTGGTGGGCGAACCCACCAAGGAAACAA
GAAGTACGCAAAAGACAAGGGTGAATAATTCAGCAAGGCTGTAACCCCGAGAGGTTCCAGCCCGCCTTCATATTTGTGTA
ATGATCCCTCCGCAGGTTCACCTACGGAGACCTTGTTACGACTTTTACTTCCTCTAAATGACCAAG

Sorry it might not have been clear before hand, but I am looking ot get something like this:

>barcodelabel=#ITS2_A_B1_;size=8857;
TTAATTTGTTACTGACGCTGATTGCAATTACAAAAGGTTTATGTTTGTCCTAGTGGTGGGCGAACCCACCAAGGAAACAA
GAAGTACGCAAAAGACAAGGGTGAATAATTCAGCAAGGCTGTAACCCCGAGAGGTTCCAGCCCGCCTTCATATTTGTGTA
ATGATCCCTCCGCAGGTTCACCTACGGAGACCTTGTTACGACTTTTACTTCCTCTAAATGACCAAG

[iraun]

Yep, sorry, I understood the opposite, see RamRS comment, he has written the correct command for your goal.