Install Alpha Fold 2.0 Protein structure prediction

Dear All
I wonder if any Body as succeed to Install Alpha Fold 2.= or a similar program on An HPC Clusters in Geneva
Best regards
Olivier Schaad

Dear Olivier,

I have a Singularity installation of Colabfold on Baobab, but it is unusable because of very slow IO speed during sequence alignment. This step which is suppose to last few minutes, last several hours.

So I would also be interested also if someone is able to have something usable.

Best regards,
Julien

Dear HPC team,

I just saw this announcement of the installation of AlphaFold on Baobab and was able to load the module with the command:

ml GCC/10.2.0  CUDA/11.1.1  OpenMPI/4.0.5 AlphaFold/2.1.2-TensorFlow-2.5.0

I’m now looking at the location of the ~2Tb database that is required to run alphafold (see here:GitHub - google-deepmind/alphafold: Open source code for AlphaFold.). Could you please indicate where the database has been installed, and ideally provide a test command line that can be used to run Alphafold on the HPC cluster.

Thanks a lot for your help
Julien

Dear Julien,

We have tried to host database on a fast storage shared across all node. Could you please check the database in the folder “/srv/fast/alphafold”.

Thanks for help,