Alphafold2 V2.0安装

源文件所在：

非docker安装：

cuda版本11.3才能按照给的配置安装，不然会报错！！！

因为cudatoolskit cudnn tensorflow jaxlib要和cuda版本对应！！！

默认的参数运行很慢，可修改一下文件夹中py文件的n_cpu, maxseq, realign_max, maxfit参数进行加速：

/home/databank/alphafold2/alphafold/alphafold/data/tools/

运行参数说明：

Usage: run_alphafold.sh 

Required Parameters:

-d      Path to directory of supporting data

-o    Path to a directory that will store the results.

-m   Names of models to use (a comma separated list)

-f    Path to a FASTA file containing one sequence

-t  Maximum template release date to consider (ISO-8601 format - i.e. YYYY-MM-DD). Important if folding historical test sets

Optional Parameters:

-b     Run multiple JAX model evaluations to obtain a timing that excludes the compilation time, which should be more indicative of the time required for inferencing many proteins (default: 'False')

-g       Enable NVIDIA runtime to run with GPUs (default: 'True')

-a   Comma separated list of devices to pass to 'CUDA_VISIBLE_DEVICES' (default: 'all')

-p        Choose preset model configuration - no ensembling (full_dbs) or 8 model ensemblings (casp14) (default: 'full_dbs')

执行命令：

bash run_alphafold.sh -d /home/databank/alphafold2/alphafold/Genetic_databases/ -o ./example/ -m model_1,model_2,model_3,model_4,model_5 -f ./example/query.fasta -t 2020-05-14 -a 1

制定日期-t   搜索模板时是往过去搜索的！！！

nohup bash run_alphafold.sh -d /home/databank/alphafold2/alphafold/Genetic_databases/ -o ./example/ -m model_1,model_2,model_3,model_4,model_5 -f ./example/query.fasta -t 2020-05-14 -a 0 >> run.log 2>&1 &

输出重定向到log文件

complex_prediction

有时候到hhblits这一步会卡住，可尝试调整所用cpu数，单独运行那一步看是否能够正常运行。