What's the location of the trained model when generating a sample from given MSA

[psp3dcg]

Thanks for the really nice package! A question (rather than an issue): is there any official trained model and where does it exist in package? Thank U :)

[pagnani]

Hi @psp3dcg. Sorry for the late reply. We do not save any "officially trained model", particularly, because the training is very fast. How large is the dataset of interest for you?

[psp3dcg]

OK, thank you for the reply~

[pagnani]

Hi @psp3dcg, I saw you reopened this. Anything else I can do here?

[psp3dcg]

Hi @psp3dcg, I saw you reopened this. Anything else I can do here?

Oh, yes. I want to ask the name of protein families you used during the training except for PF00072 and PF00196, and the number of sequences in each family. Thank you :)

[pagnani]

We used ArDCA for many protein families. In the Readme of the package we mention a companion repo ArDCAData with data from 5 protein families (PF00014, PF00072, PF00076, PF00595, PF13354). If you need more, MSA for all protein families can be found Interpro.