search

microsoft / evodiff

Generation of protein sequences and evolutionary alignments via discrete diffusion models
MIT License
516 stars 73 forks source link

Question: ways to improve conditional MSA generation #33

Closed yzhang-github-pub closed 3 months ago

yzhang-github-pub commented 8 months ago

I played with evo-diff on msa conditional generation. Structures of query and generated are not as similar as I expected. Actually many times they are quite different.

Below is a comparison of 3D structures of query and generated sequence from your example jupyter notebook under "Evolutionary guided sequence generation with EvoDiff-MSA":

query=MNIFEMLRIDEGLRLKIYKDTEGYYTIGIGHLLTKSPSLNAAKSELDKAIGRNTNGVITKDEAEKLFNQDVDAAVRGILRNAKLKPVYDSLDAVRRAALINMVFQMGETGVAGFTNSLRMLQQKRWDEASVNLAKSRWYNQTPNRAKRVITTFRTGTWDAYKNL generated=MDLRSSLVEHEGLRWKVYNNAEYVPTIGLGQIHNRPSQYWDYPVPLPEQYAEKDQISWSLETIQAVFDERYTKAKSEMVNLETIGKNFDDLPSEHTNAVTDMMFQLGTDHLSEFHKMITALKNNTYEEACREMKSSFWTRQMGNRCTRYLNDALEENYFFFNHH

Structures are from colab alphafold2 with default settings. Arrows pointing to regions where structures are not similar: query_vs_gen

My question is how to generate sequences with more similar structures to that of query. Should input a3m include very similar sequences? Any parameters to test in generation, like number of sequences to subsample? Thanks.

yangkky commented 3 months ago

Some things to try: