FreshAirTonight / af2complex

Predicting direct protein-protein interactions with AlphaFold deep learning neural network models.
146 stars 19 forks source link

error on installing AF2complex on colab notebook #23

Open herman-einstein opened 9 months ago

herman-einstein commented 9 months ago

When I try to run AF2complex on colab, I come across the following error. I will very much appreciate if you can solve this error for me.

螢幕擷取畫面 2023-12-08 001843

FreshAirTonight commented 9 months ago

The reason for this error is because Colab updated their environment, which unforuntaly is incompatible with the current notebook. If you have direct access to a GPU node, you could try install AF2Complex and run it directly.

herman-einstein commented 9 months ago

I see. Thank you very much

herman-einstein commented 8 months ago

I am using school's server to run alphafold and command is somthing like the following

_module add singularity/3.10.0 SIMG=/share/apps/singularity/simg/alphafold/alphafold_2_3_1.sif singularity exec --nv --bind /data:/data,/share:/share,/home:/home $SIMG bash -c "ulimit -s unlimited; python /app/alphafold/run_alphafold.py --fasta_paths=/home/alphafold_analysis/output/query.fasta --max_template_date=2023-12-01 --model_preset=multimer --data_dir=/home/alphafold_analysis/alphafold_database --output_dir=/home/alphafold_analysis/output --uniref90_database_path=/home/alphafold_analysis/alphafold_database/uniref90/uniref90.fasta --mgnify_database_path=/home/alphafold_analysis/alphafold_database/mgnify/mgy_clusters_2022_05.fa --template_mmcif_dir=/home/alphafold_analysis/alphafold_database/pdb_mmcif/mmcif_files --obsolete_pdbs_path=/home/alphafold_analysis/alphafold_database/pdb_mmcif/obsolete.dat --pdb_seqres_database_path=/home/alphafold_analysis/alphafold_database/pdb_seqres/pdb_seqres.txt --uniprot_database_path=/home/alphafold_analysis/alphafold_database/uniprot/uniprot.fasta --db_preset=reduced_dbs --small_bfd_database_path=/home/alphafold_analysis/alphafold_database/small_bfd/bfd-first_non_consensus_sequences.fasta --use_gpurelax=False"

I only have limited access to Alphafold using the server. Moreover, I think I am using singularity to run Alphafold instead of docker. May I know where I should alter the command so that I can run af2complex propoerly?

herman-einstein commented 8 months ago

May I know what version of jaxlib I should have if I install jax==0.3.25 because I following the command you suggested

_CUDA=11.1.1 pip3 install --upgrade --no-cache-dir jax==0.2.14 \ jaxlib==0.1.69+cuda$(cut -f1,2 -d. <<< ${CUDA} | sed 's/.//g') \ -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html_

but there is other error like image

I very much appreciate if you could suggest me the version of installing jax and jaxlib

FreshAirTonight commented 8 months ago

Please try:

cd af2complex directory
$ pip3 install -r requirements.txt
$ pip3 install jax==0.3.25 jaxlib==0.3.25+cuda11.cudnn805 -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html
herman-einstein commented 8 months ago

Please try:

cd af2complex directory
$ pip3 install -r requirements.txt
$ pip3 install jax==0.3.25 jaxlib==0.3.25+cuda11.cudnn805 -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html

Very much appreciate your help.

In addition, I find that scipy.spatial is needed in your modified alphafold script "confidence.py", line 532 but you did not import this package for the command.

Again, thank you very much for your help.