Parameters for EVCouplings Jackhmmer search

[phatma...@gmail.com]

I'm trying to understand how evfold generates its set of sequences consulting either HHblits or jackhmmer.

If I do a default search on hhblits, or an jackhmmer, using E-value of -3, and with all other parameters default, I don't get the same set of sequences out that is generated as the output alignment from evfold.

For my protein (3hjl), ev couplings running through jackhmmer gets 4055 sequences in an MSA file. Running jackhmmer for the same protein sequence, returns only 1766 sequences.

Am I missing something in how the sets of sequences are recalled from either hhblits or jackhmmer? If anything I would have expected less as a default jackhmmer search throws up many redundant entries.

Thanks,
Matt.