FASTA with multiple entries

[fjell-dev]

Hi, I would like to ask how SimpleFold processes FASTA files with multiple entries. It looks like that it will only take the first entry in the input file. Do I need to split the input file to multiple single-entry files? Thank you.

[durubing-jn]

Hi,
I have same question, may I ask if you have solved it?

[mbautistamartin]

Good question! What would be the intended use for having a FASTA file with multiple entries? Note that SimpleFold was not trained with multimer or protein complex data so we have not enabled that kind of use case yet. If the goal is to generate structures for different independent sequences I'd suggest to have one FASTA file per sequence.

[fjell-dev]

Hi, I have same question, may I ask if you have solved it?

You can split the input FASTA file into single ones using the split function of the seqkit toolkit (https://bioinf.shenwei.me/seqkit/usage/#split) (seqkit is very useful when dealing with sequencing data). Then you can use some one-line BASH commands to run simplefold in batch with each of the sequences. Or use fd (https://github.com/sharkdp/fd?tab=readme-ov-file) to make life easier with the -x or -X function for executing command when certain files are found (I prefer fd to the find command but the latter is normally available on UNIX-based systems).

[fjell-dev]

Good question! What would be the intended use for having a FASTA file with multiple entries? Note that SimpleFold was not trained with multimer or protein complex data so we have not enabled that kind of use case yet. If the goal is to generate structures for different independent sequences I'd suggest to have one FASTA file per sequence.

As SimpleFold does not support multimer anyway, so there should be no ambiguity when it interprets the input file? I mean, it would just make life a bit easier to have one input file containing multiple entries rather than splitting the file and writing some batch scripts. Our use case is to use SimpleFold to predict en mass sequence mining hits which can be in a range of hundreds or thousands (I know it is possible with workarounds, mostly for the convenience of integrating SimpleFold to our pipelines).

[mbautistamartin]

Thanks for the clarification! For now I think it's reasonable to have batch scripts in order to keep the call to simplefold as simple as possible.