Update README.md

and pgym official benchmark scripts

Author: niklases

.gitignore

@@ -493,3 +493,10 @@ datasets/AVGFP/MLPLMC_DE_trajectories.png
 datasets/AVGFP/Predictions_GREMLIN_TopTS.txt
 datasets/AVGFP/Predictions_PLMC_TopTS.txt
 datasets/AVGFP/uref100_avgfp_jhmmer_119_plmc_42.6.params
+
+scripts/ProteinGym_runs/official/data/cv_folds_multiples_substitutions/*
+scripts/ProteinGym_runs/official/data/cv_folds_singles_substitutions/*
+scripts/ProteinGym_runs/official/data/DMS_msa_files/*
+scripts/ProteinGym_runs/official/data/ProteinGym_AF2_structures/*
+scripts/ProteinGym_runs/official/data/DMS_substitutions.csv
+scripts/ProteinGym_runs/official/benchmark_runs/output.log

README.md

@@ -47,15 +47,6 @@ When incorporating DCA and PLM features, both models are fine-tuned via few-shot
 ## Quick Installation
 A quick installation of the PyPEF command line framework using PyPI for Linux and Windows and Python >= 3.10 can be performed with:
 
-
-<p align="center">
-  <img src=".github/imgs/splitting_workflow.png" alt="drawing" width="1000"/>
-</p>
-
-<a name="installation"></a>
-## Quick Installation
-A quick installation of the PyPEF command line framework using PyPI for Linux and Windows and Python >= 3.10 can be performed with:
-
 ```bash
 pip install -U pypef
 # optionally, for GPU support (see requirements section below):

scripts/ProteinGym_runs/official/README.md

@@ -27,4 +27,4 @@ chmod a+x ./run_over_all.sh
 ./run_over_all.sh split_method=fold_random_5
 ```
 
-All steps can also performed with running `do_all.sh` (writes output to output.log, where running `python remove_tqdm_lines_log.py` is handy for removing the tqdm progress bars from the output log for eased log reading).
+All steps can also performed with running `do_all.sh` (if writing output to output.log, running `python remove_tqdm_lines_log.py` is handy for removing the tqdm progress bars from the output log for eased log reading).

scripts/ProteinGym_runs/official/configs/proteingym_data_setup.yaml

@@ -0,0 +1,26 @@
+defaults:
+  - override hydra/job_logging: disabled  # Prevents creation of .log file from Hydra
+
+DMS_idx: ???
+split_method: ???
+llm: ???
+DMS_reference_file_path: ../data/DMS_substitutions.csv
+DMS_data_folder_singles: ../data/cv_folds_singles_substitutions
+DMS_data_folder_multiples: ../data/cv_folds_multiples_substitutions
+DMS_MSA_data_path: ../data/DMS_msa_files
+DMS_PDB_data_path: ../data/ProteinGym_AF2_structures
+output_scores_folder: model_scores/supervised_substitutions
+
+seed: 2024
+progress_bar: false
+overwrite: false
+
+# Disables Hydra logging functionality
+hydra:
+  output_subdir: null
+  job:
+    chdir: false
+  run:
+    dir: .
+
+  

scripts/ProteinGym_runs/official/do_all.sh

@@ -1,4 +1,5 @@
 #!/bin/bash
+set -e
 
 
 split_method=fold_random_5
@@ -11,4 +12,5 @@ cd ..
 ./get_py_packages.sh
 
 cd benchmark_runs
-./run_over_all.sh split_method=$split_method > output.log 2>&1 &
+./run_over_all.sh split_method=$split_method # > output.log 2>&1 &
+

scripts/ProteinGym_runs/official/get_py_packages.sh

@@ -1,7 +1,9 @@
 #!/bin/bash
 
-gh repo clone niklases/pypef
-cd PyPEF
+wget https://github.com/niklases/PyPEF/archive/refs/heads/main.zip
+unzip main.zip && rm main.zip
+cd PyPEF-main/
 pip install -r requirements.txt
 pip install hydra-core
 pip install .
+cd ..