Removal of sequences that is partial of a longer sequences in multifasta file

Question

[Louis Kok]

Hi. I wish to remove any sequence that is partial of a longer sequence in multifasta file. For example, let say I have three sequences below:

>seq1
ACGACGATCGT**ACTAGCATCGAGCGTAC**TACGTAGCGCGT

>seq2
**ACTAGCATCGAGCGTAC**

>seq3
AGCAGCGTACGTGACTACGACGATCTACGTATCTAGCTCGTACACT

seq2 is exactly part of seq1. So after removing the partial (duplicate) sequences, I am expecting to have the following multifasta file:

>seq1
ACGACGATCGTACTAGCATCGAGCGTACTACGTAGCGCGT

>seq3
AGCAGCGTACGTGACTACGACGATCTACGTATCTAGCTCGTACACT

All the answers I managed to search are removal of exact duplicates. Is there any tool or script to achieve the purpose? Thanks in advance.

[Chirag Parsania]

You can try program CD-HIT with Sequence Identity Parameter = 1. It will cluster all sequences which are identical and return you longest one for each cluster.

Answer 1

[Ram]

I think the most plain way would be to write a custom script using BioPython. You could create a dict with the identifiers as key and sequence as value, then test if each sequence, starting from the smallest one, is a substring of larger sequences. You can use that to pick relevant sequences and save them to an output file.