Changes between Version 12 and Version 13 of FAQ

Timestamp:

11/02/21 12:25:48 (3 years ago)

Author:

gbell

Comment:

--

FAQ

@@ -210 +210 @@
 1. [=#alphafold How can I run '''AlphaFold 2.0''' here at Whitehead?]
   * You'll need access to a system with a GPU, or the new GPU queue.  IT can help you choose and obtain access to GPU systems if required.
-  * Modify the environment with a path to the required data:
-    * export ALPHAFOLD_DATA_PATH=/alphafold/data
-  * [Optional] Get information about the 'alphafold' command:
+  * For a working command to be executed on fry.wi.mit.edu, which then sends the AlphaFold command to the GPU node on the slurm cluster, see /nfs/BaRC_Public/BaRC_code/shell/run_AlphaFold/Commands.sh
+  * Start by copying RunAlphaFold_slurm.sh to your project directory.
+  * The main command looks like
 {{{
-singularity run -B $ALPHAFOLD_DATA_PATH:/data -B .:/etc \
---pwd /app/alphafold \
---nv /alphafold/alphafold_2.0.0.sif --helpshort
+sbatch -J AF_1 --export=ALL,FASTA_NAME=Sample_protein_1.fa,USERNAME=myUsername,FASTA_PATH=proteins,AF2_WORK_DIR=/nfs/BaRC_Public/BaRC_code/shell/run_AlphaFold ./RunAlphaFold_slurm.sh
 }}}
-  * Run the main AlphaFold 2.0 command (replacing query.fasta with your input protein sequence), like
-   {{{
-singularity run -B $ALPHAFOLD_DATA_PATH:/data -B .:/etc \
---pwd /app/alphafold \
---nv /alphafold/alphafold_2.0.0.sif \
---fasta_paths=/data/query.fasta \
---output_dir=/tmp/AlphaFold_output \
---model_names=model_1 \
---preset=full_dbs \
---data_dir=/data \
---max_template_date=2020-05-14 \
---uniref90_database_path=/data/uniref90/uniref90.fasta \
---mgnify_database_path=/data/mgnify/mgy_clusters.fa \
---uniclust30_database_path=/data/uniclust30/uniclust30_2018_08/uniclust30_2018_08 \
---bfd_database_path=/data/bfd/bfd_metaclust_clu_complete_id30_c90_final_seq.sorted_opt \
---pdb70_database_path=/data/pdb70/pdb70 \
---template_mmcif_dir=/data/pdb_mmcif/mmcif_files \
---obsolete_pdbs_path=/data/pdb_mmcif/obsolete.dat
-}}}
-
+    where the inputs are
+    * AF2_WORK_DIR => project working directory
+    * FASTA_PATH   => directory within AF2_WORK_DIR with input protein sequence
+    * FASTA_NAME   => name of input protein sequence file within $AF2_WORK_DIR/$FASTA_PATH
+    * USERNAME     => Username for job submission and email
   * More information, including an explanation of the output files, is here: https://github.com/deepmind/alphafold