参考:
https://github.com/facebookresearch/esm
https://huggingface.co/facebook/esm2_t33_650M_UR50D
https://esmatlas.com/resources?action=fold
硬件运行加速版:
https://github.com/pengzhangzhi/faesm
新版本:
ESM C、ESM 3
https://github.com/evolutionaryscale/esm
方法1、transformers版本使用
直接输入Fasta 氨基酸序列格式就行;第一次下载esm2_t33_650M_UR50D模型有点慢,有2个多G大
from transformers import BertTokenizer, AutoTokenizer,BertModel, AutoModel
tokenizer_ = AutoTokenizer.from_