add slurm scripts, update sh script

.gitignore

@@ -1 +0,0 @@
-*.slurm

alphafold/model/data.py

@@ -22,10 +22,10 @@
 # Internal import (7716).
 
 
-def get_model_haiku_params(model_name: str, data_dir: str) -> hk.Params:
+def get_model_haiku_params(model_name: str, parameter_path: str) -> hk.Params:
   """Get the Haiku parameters from a model name."""
 
-  path = os.path.join(data_dir, 'params', f'params_{model_name}.npz')
+  path = os.path.join(parameter_path, f'params_{model_name}.npz')
 
   with open(path, 'rb') as f:
     params = np.load(io.BytesIO(f.read()), allow_pickle=False)

run_alphafold.py

@@ -50,7 +50,7 @@
     'separated by commas. All FASTA paths must have a unique basename as the '
     'basename is used to name the output directories for each prediction.')
 flags.DEFINE_list('model_names', None, 'Names of models to use.')
-flags.DEFINE_string('data_dir', None, 'Path to directory of supporting data.')
+flags.DEFINE_string('parameter_path', None, 'Path to directory of supporting data.')
 flags.DEFINE_string('output_dir', None, 'Path to a directory that will '
                     'store the results.')
 flags.DEFINE_string('jackhmmer_binary_path', shutil.which('jackhmmer'),
@@ -389,7 +389,7 @@ def main(argv):
       model_config.model.num_recycle = FLAGS.recycling
       model_config.data.common.num_recycle = FLAGS.recycling
     model_params = data.get_model_haiku_params(
-        model_name=model_name, data_dir=FLAGS.data_dir)
+        model_name=model_name, parameter_path=FLAGS.parameter_path)
     model_runner = model.RunModel(model_config, model_params)
     for i in range(num_predictions_per_model):
       model_runners[f'{model_name}_pred_{i}'] = model_runner
@@ -433,7 +433,7 @@ def main(argv):
   flags.mark_flags_as_required([
       'fasta_paths',
       'output_dir',
-      'data_dir',
+      'parameter_path',
       'uniref90_database_path',
       'mgnify_database_path',
       'template_mmcif_dir',

run_alphafold.sh

@@ -170,6 +170,7 @@ export TF_FORCE_UNIFIED_MEMORY='1'
 export XLA_PYTHON_CLIENT_MEM_FRACTION='4.0'
 
 # Path and user config (change me if required)
+parameter_path="$data_dir/params/2.2"
 bfd_database_path="$data_dir/bfd/bfd_metaclust_clu_complete_id30_c90_final_seq.sorted_opt"
 small_bfd_database_path="$data_dir/small_bfd/bfd-first_non_consensus_sequences.fasta"
 mgnify_database_path="$data_dir/mgnify/mgy_clusters_2018_12.fa"
@@ -214,7 +215,7 @@ fi
 python $alphafold_script \
 --fasta_paths=$fasta_path \
 --model_names=$model_selection \
---data_dir=$data_dir \
+--parameter_path=$parameter_path \
 --output_dir=$output_dir \
 --jackhmmer_binary_path=$jackhmmer_binary_path \
 --hhblits_binary_path=$hhblits_binary_path \

sub_alphafold.slurm

@@ -0,0 +1,24 @@
+#!/bin/bash
+#SBATCH --job-name=GA98
+#SBATCH --partition=a100
+#SBATCH -N 1
+#SBATCH --ntasks-per-node=1
+#SBATCH --cpus-per-task=8
+#SBATCH --gres=gpu:1
+#SBATCH --output=task_file/%x_%j.out
+#SBATCH --error=task_file/%x_%j.err
+
+module purge
+module load miniconda3
+source activate parafold
+
+./run_alphafold.sh \
+-d data \
+-o output \
+-p monomer_ptm \
+-u 0 \
+-i \
+input/mono_set_1/GA98.fasta \
+-t 2020-01-01 \
+-r best 
+

sub_alphafold_multimer.slurm

@@ -0,0 +1,25 @@
+#!/bin/bash
+#SBATCH --job-name=complex_benchmark
+#SBATCH --partition=a100
+#SBATCH -N 1
+#SBATCH --ntasks-per-node=1
+#SBATCH --cpus-per-task=8
+#SBATCH --gres=gpu:2
+#SBATCH --output=task_file/%x_%j.out
+#SBATCH --error=task_file/%x_%j.err
+
+module purge
+module load miniconda3
+source activate parafold
+
+./run_alphafold.sh \
+-d data \
+-o output \
+-p multimer \
+-m model_1_multimer_v3,model_2_multimer_v3,model_3_multimer_v3,model_4_multimer_v3,model_5_multimer_v3 \
+-u 0,1 \
+-i \
+input/complex_benchmark/MHC_TCR.fasta \
+-t 2022-01-01 \
+-r none
+

tools/batch_scripts/template_alphafold.slurm

@@ -0,0 +1,17 @@
+#!/bin/bash
+#SBATCH --job-name=A501_001
+#SBATCH --partition=dgx2
+##SBATCH -w vol08
+#SBATCH -N 1
+#SBATCH --ntasks-per-node=1
+#SBATCH --cpus-per-task=6
+#SBATCH --gres=gpu:1
+#SBATCH --output=task_file/%x_%j.out
+#SBATCH --error=task_file/%x_%j.err
+
+module load cuda/10.1.243-gcc-8.3.0
+module load miniconda3
+source activate alphafold_cuda10
+cd /lustre/home/acct-stu/stu/deepocean
+/lustre/home/acct-stu/stu/parallelfold/run_alphafold.sh -d /lustre/home/acct-stu/stu/parallelfold/data -o A501_result -m model_1,model_2 -f A501_seq/${SLURM_JOB_NAME}.fasta -t 2021-07-27
+

tools/batch_scripts/template_feature.slurm

@@ -0,0 +1,17 @@
+#!/bin/bash
+#SBATCH --job-name=A501_001
+#SBATCH --partition=small
+#SBATCH -N 1
+#SBATCH -n 8
+#SBATCH --ntasks-per-node=8
+#SBATCH --output=task_file/%x_%j.out
+#SBATCH --error=task_file/%x_%j.err
+
+echo $SLURM_JOB_NAME
+
+module load cuda/10.1.243-gcc-8.3.0
+module load miniconda3
+source activate alphafold_cuda10
+cd /lustre/home/acct-stu/stu/deepocean
+
+/lustre/home/acct-stu/stu/parallelfold/run_feature.sh -d /lustre/home/acct-stu/stu/parallelfold/data -o A501_result -m model_1 -f A501_seq/${SLURM_JOB_NAME}.fasta -t 2021-07-27