Datasets:
File size: 6,308 Bytes
84a242a | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 | #!/bin/bash
export NCCL_ASYNC_ERROR_HANDLING=1
export TORCH_NCCL_BLOCKING_WAIT=1
export TORCH_NCCL_HEARTBEAT_TIMEOUT_SEC=7200 # or longer
export TORCH_NCCL_ENABLE_MONITORING=0
#----------------------------------------Model and Benchmark Settings----------------------------------------#
# Devices to use
devices="0 1 2 3 4 5 6 7"
# '0' for training, '1' for testing
test=1
# For test, lightning saved version
# The first of the 5-fold versions to test, e.g., if trained versions are 20,21,22,23,24, set version=20
version=5
# (optional) a previous 5-fold result for comparison
original_version=90
# Benchmark to train/test; choose from:
# benchmark='posex'
# benchmark="astex_posex"
benchmark="moad"
# benchmark="posex_self_docking"
# benchmark="posex_cross_docking"
# benchmark='astex'
# benchmark='posebusters'
# Relaxation setting for the benchmark, 'both' for including both relaxed and unrelaxed poses
relaxation='both' # 'both', 'true', or 'false'
# Model to train
model='MolAS'
# Jump per-fold evaluation and use saved results (can activate after testing once)
jump_eval='false'
seed=42
#------------------------------------------------------------------------------------------------------------#
num_devices=$(echo $devices | wc -w)
declare -A num_classes_both_dict=(
[posex]=48
[astex]=48
[astex_posex]=48
[moad]=8
[posex_self_docking]=48
[posex_cross_docking]=48
[posebusters]=14
[posebench]=14
)
declare -A num_classes_dict=(
[posex]=24
[astex]=24
[astex_posex]=24
[moad]=8
[posex_self_docking]=24
[posex_cross_docking]=24
[posebusters]=7
[posebench]=7
)
declare -A incl_columns_dict=(
[posex]='surfdock unimol diffdock diffdock_l gnina autodock glide alphafold3 moe dynamicbind rfaa fabind equibind boltz1x tankbind neuralplexer chai deepdock boltz ifd protenix diffdock_pocket DS interformer'
[astex]='surfdock unimol diffdock diffdock_l gnina autodock glide alphafold3 moe dynamicbind rfaa fabind equibind boltz1x tankbind neuralplexer chai deepdock boltz ifd protenix diffdock_pocket DS interformer'
[astex_posex]='surfdock unimol diffdock diffdock_l gnina autodock glide alphafold3 moe dynamicbind rfaa fabind equibind boltz1x tankbind neuralplexer chai deepdock boltz ifd protenix diffdock_pocket DS interformer'
[moad]=''
[posex_self_docking]='surfdock unimol diffdock diffdock_l gnina autodock glide alphafold3 moe dynamicbind rfaa fabind equibind boltz1x tankbind neuralplexer chai deepdock boltz ifd protenix diffdock_pocket DS interformer'
[posex_cross_docking]='surfdock unimol diffdock diffdock_l gnina autodock glide alphafold3 moe dynamicbind rfaa fabind equibind boltz1x tankbind neuralplexer chai deepdock boltz ifd protenix diffdock_pocket DS interformer'
[posebusters]='deepdock diffdock equibind gold tankbind unimol autodock'
)
echo "Number of classes for benchmark $benchmark: $num_classes"
if [ "$relaxation" == "both" ]; then
incl_columns=''
num_classes=${num_classes_both_dict[$benchmark]}
elif [ "$relaxation" == "true" ]; then
# Add '_relaxation' suffix to each method in incl_columns
# incl_columns="${incl_columns_dict[$benchmark]}"_relaxation only adds suffix to the last method
incl_columns=""
for method in ${incl_columns_dict[$benchmark]}; do
incl_columns+="${method}_relaxation "
done
echo "Included columns with relaxation: $incl_columns"
num_classes=${num_classes_dict[$benchmark]}
else
incl_columns="${incl_columns_dict[$benchmark]}"
num_classes=${num_classes_dict[$benchmark]}
fi
log_root="lightning_logs"
log_name="benchmark_experiment_${benchmark}"
if [ "$model" == "MCGNNASDock" ]; then
num_node_features=960
else
num_node_features=1152
fi
if [ "$test" -eq 0 ]; then
for i in {0..4}; do
echo "Training with fold $i"
torchrun --standalone --nproc_per_node=$num_devices main.py \
--k_folds 5 \
--fold_num $i \
--num_node_features $num_node_features \
--max_epochs 300 \
--devices $devices \
--num_classes $num_classes \
--incl_columns $incl_columns \
--batch_size 16 \
--num_workers 0 \
--lr 1e-3 \
--benchmark $benchmark \
--seed $seed \
--test_size 0.2 \
--log_root $log_root \
--log_name $log_name \
--model $model
done
else
original_ckpts=()
after_ckpts=()
for i in {0..4}; do
echo "Testing with fold $i"
# Detect .ckpt files starting with 'best-epoch=xxx' with largest xxx
dir="${log_root}/${log_name}/version_${version}/checkpoints"
original_dir="${log_root}/${log_name}/version_${original_version}/checkpoints"
# Find all files matching the pattern
best_ckpt=$(find "$dir" -type f -name 'best-epoch=*-*.ckpt' | \
grep -oE 'best-epoch=([0-9]+)-[^/]*\.ckpt$' | \
sed -E 's/^best-epoch=([0-9]+)-.*/\1 &/' | \
sort -nr | \
head -n1 | \
cut -d' ' -f2-)
original_best_ckpt=$(find "$original_dir" -type f -name 'best-epoch=*-*.ckpt' | \
grep -oE 'best-epoch=([0-9]+)-[^/]*\.ckpt$' | \
sed -E 's/^best-epoch=([0-9]+)-.*/\1 &/' | \
sort -nr | \
head -n1 | \
cut -d' ' -f2-)
best_ckpt_path="$dir/$best_ckpt"
original_best_ckpt_path="${log_root}/benchmark_experiment_astex_posex/version_${original_version}/checkpoints/$original_best_ckpt"
original_ckpts+=("$original_best_ckpt_path")
after_ckpts+=("$best_ckpt_path")
# jump_eval='true' to enable jump evaluation
if [ "$jump_eval" == "false" ]; then
echo "#---------------------------Testing with checkpoint: $best_ckpt_path---------------------------#"
python main.py \
--k_folds 5 \
--fold_num $i \
--num_node_features $num_node_features \
--devices $devices \
--num_classes $num_classes \
--incl_columns $incl_columns \
--num_workers 1 \
--benchmark $benchmark \
--seed $seed \
--test_size 0.2 \
--test \
--ckpt_path $best_ckpt_path \
--model $model
echo "Finished testing with checkpoint: $best_ckpt_path"
fi
version=$((version + 1))
original_version=$((original_version + 1))
done
python stats_test.py \
--original_ckpts "${original_ckpts[@]}" \
--after_ckpts "${after_ckpts[@]}"
fi |