hpcaitech / FastFold

Optimizing AlphaFold Training and Inference on GPU Clusters
Apache License 2.0
566 stars 86 forks source link

modify reading data way and add inference test #50

Closed Gy-Lu closed 2 years ago

Gy-Lu commented 2 years ago

By modifying the way of reading data, we can now process sequences with \n. Meanwhile, it supports multi-sequences. Below is my test fasta:

 >test
 ANKASDKHASJHFJKA
 ASJKDFAKJSHDJKJ
 >test2
 FADHJKDEASHFKLKLD
 >test3
 DHLFLAJKJKLAJKLFKLASJKL

And the code returns tags and seqs:

image
Gy-Lu commented 2 years ago

By the way, I added some unit test for inference, which uses random data to test if the code could run correctly.

Fazziekey commented 2 years ago

the branch maybe should merge into branch develop rather than main

Gy-Lu commented 2 years ago

the branch maybe should merge into branch develop rather than main

No, I think main is fine. For this pr could be seen as bug fixing rather than feature adding. Besides, develop branch would be deleted after merging into main.