[](https://github.com/pgmikhael/CLIPZyme/blob/main/LICENSE.txt)
[](https://arxiv.org/abs/2402.06748)
<!--  -->
# CLIPZyme
Implementation of the paper [**CLIPZyme: Reaction-Conditioned Virtual Screening of Enzymes**](https://github.com/pgmikhael/CLIPZyme/blob/main/LICENSE.txt)
Table of contents
=================
<!--ts-->
- [CLIPZyme](#clipzyme)
- [Table of contents](#table-of-contents)
- [Installation:](#installation)
- [Checkpoints and Data Files:](#checkpoints-and-data-files)
- [Screening with CLIPZyme](#screening-with-clipzyme)
- [Using CLIPZyme's screening set](#using-clipzymes-screening-set)
- [Using your own screening set](#using-your-own-screening-set)
- [Interactive (slow)](#interactive-slow)
- [Batched (fast)](#batched-fast)
- [Reproducing published results](#reproducing-published-results)
- [Data processing](#data-processing)
- [Training and evaluation](#training-and-evaluation)
- [Downloading Batched AlphaFold Database Structures](#downloading-batched-alphafold-database-structures)
- [Citation](#citation)
<!--te-->
# Installation:
1. Clone the repository:
```bash
git clone https://github.com/pgmikhael/CLIPZyme.git
```
2. Install the dependencies:
```bash
cd clipzyme
conda env create -f environment.yml
conda activate clipzyme
python -m pip install clipzyme
```
3. Download ESM-2 checkpoint `esm2_t33_650M_UR50D`. The `esm_dir` argument should point to this directory. The following command will download the checkpoint directly:
```bash
wget https://dl.fbaipublicfiles.com/fair-esm/models/esm2_t33_650M_UR50D.pt
```
# Checkpoints and Data Files:
The model checkpoint and data are available on Zenodo [here](https://zenodo.org/records/11187895):
- [clipzyme_data.zip](https://zenodo.org/records/11187895/files/clipzyme_data.zip?download=1):
- The following commands will download the checkpoint directly:
```
wget https://zenodo.org/records/11187895/files/clipzyme_data.zip
unzip clipzyme_data.zip -d files
```
- Note that the data files should be extracted into the `files/` directory.
- `enzymemap.json`: contains the EnzymeMap dataset.
- `cached_enzymemap.p`: contains the processed EnzymeMap dataset.
- `clipzyme_screening_set.p`: contains the screening set as dict of UniProt IDs and precomputed protein embeddings.
- `uniprot2sequence.p`: contains the mapping form sequence ID to amino acids.
- [clipzyme_model.zip](https://zenodo.org/records/11187895/files/clipzyme_model.zip?download=1):
- The following command will download the checkpoint directly:
```
wget https://zenodo.org/records/11187895/files/clipzyme_model.zip
unzip clipzyme_model.zip -d files
```
- `clipzyme_model.ckpt`: the trained model checkpoint.
# Screening with CLIPZyme
## Using CLIPZyme's screening set
First, download the screening set and extract the files into `files/`.
```python
import pickle
from clipzyme import CLIPZyme
## Load the screening set
##-----------------------
screenset = pickle.load(open("files/clipzyme_screening_set.p", 'rb'))
screen_hiddens = screenset["hiddens"] # hidden representations (261907, 1280)
screen_unis = screenset["uniprots"] # uniprot ids (261907,)
## Load the model and obtain the hidden representations of a reaction
##-------------------------------------------------------------------
model = CLIPZyme(checkpoint_path="files/clipzyme_model.ckpt")
reaction = "[CH3:1][N+:2]([CH3:3])([CH3:4])[CH2:5][CH:6]=[O:7].[O:9]=[O:10].[OH2:8]>>[CH3:1][N+:2]([CH3:3])([CH3:4])[CH2:5][C:6](=[O:7])[OH:8].[OH:9][OH:10]"
reaction_embedding = model.extract_reaction_features(reaction=reaction) # (1,1280)
enzyme_scores = screen_hiddens @ reaction_embedding.T # (261907, 1)
```
## Using your own screening set
Prepare your data as a CSV in the following format, and save it as `files/new_data.csv`. For the cases where we wish only to obtain the hidden representations of the sequences, the `reaction` column can be left empty (and vice versa).
| reaction | sequence | protein_id | cif |
| ------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------- | -------- |
| [CH3:1][N+:2]([CH3:3])([CH3:4])[CH2:5][CH:6]=[O:7].[O:9]=[O:10].[OH2:8]>>[CH3:1][N+:2]([CH3:3])([CH3:4])[CH2:5][C:6](=[O:7])[OH:8].[OH:9][OH:10] | MGLSDGEWQLVLNVWGKVEAD<br>IPGHGQEVLIRLFKGHPETLE<br>KFDKFKHLKSEDEMKASEDLK<br>KHGATVLTALGGILKKKGHHE<br>AELKPLAQSHATKHKIPIKYL<br>EFISEAIIHVLHSRHPGDFGA<br>DAQGAMNKALELFRKDIAAKY<br>KELGYQG | P69905 | 1a0s.cif |
### Interactive (slow)
```python
from torch.utils.data import DataLoader
from clipzyme import CLIPZyme
from clipzyme import ReactionDataset
from clipzyme.utils.loading import ignore_None_collate
## Create reaction dataset
#-------------------------
reaction_dataset = DataLoader(
ReactionDataset(
dataset_file_path = "files/new_data.csv",
esm_dir = "/path/to/esm2_dir",
protein_cache_dir = "/path/to/protein_cache", # optional, where to cache processed protein graphs
),
batch_size=1,
collate_fn=ignore_None_collate,
)
## Load the model
#----------------
model = CLIPZyme(checkpoint_path="files/clipzyme_model.ckpt")
model = model.eval() # optional
## For reaction-enzyme pair
#--------------------------
for batch in reaction_dataset:
output = model(batch)
enzyme_scores = output.scores
protein_hiddens = output.protein_hiddens
reaction_hiddens = output.reaction_hiddens
## For sequences only
#--------------------
for batch in reaction_dataset:
protein_hiddens = model.extract_protein_features(batch)
## For reactions only
#--------------------
for batch in reaction_dataset:
reaction_hiddens = model.extract_reaction_features(batch)
```
### Batched (fast)
1. Update the screening config `configs/screening.json` with the path to your data and indicate what you want to save and where:
```JSON
{
"dataset_file_path": ["files/new_data.csv"],
"inference_dir": ["/where/to/save/embeddings_and_scores"],
"save_hiddens": [true], # whether to save the hidden representations
"save_predictions": [true], # whether to save the reaction-enzyme pair scores
"use_as_protein_encoder": [true], # whether to use the model as a protein encoder only
"use_as_reaction_encoder": [true], # whether to use the model as a reaction encoder only
"esm_dir": ["/data/esm/checkpoints"], path to ESM-2 checkpoints
"gpus": [8], # number of gpus to use,
"protein_cache_dir": ["/path/to/protein_cache"], # where to save the protein cache [optional]
...
}
```
If you want to use specific GPUs, you can specify them in the `available_gpus` field. For example, to use GPUs 0, 1, and 2, set `available_gpus` to `["0,1,2"]`.
2. Run the dispatcher with the screening config:
```bash
python scripts/dispatcher.py -c configs/screening.json -l ./logs/
```
3. Load the saved embeddings and scores:
```python
from clipzyme import collect_screening_results
screen_hiddens, screen_unis, enzyme_scores = collect_screening_results("configs/screening.json")
```
---------------------
# Reproducing published results
## Data processing
We obtain the data from the following sources:
- [EnzymeMap:](`https://doi.org/10.5281/zenodo.7841780`) Heid et al. Enzymemap: Curation, validation and data-driven prediction of enzymatic reactions. 2023.
- [Terpene Synthases:](`https://zenodo.org/records/10567437`) Samusevich et al. Discovery and characterization of terpene synthases powered by machine learning. 2024.
Our processed data is can be downloaded from [here](https://zenodo.org/records/11187895).
## Training and evaluation
1. To train the models presented in the tables below, run the following command:
```
python scripts/dispatcher.py -c {config_path} -l {log_path}
```
- `{config_path}` is the path to the config file in the table below
- `{log_path}` is the path in which to save the log file.
For example, to run the first row in Table 1, run:
```
python scripts/dispatcher.py -c configs/train/clip_egnn.json -l ./logs/
```
2. Once you've trained the model, run the eval config to evaluate the model on the test set. For example, to evaluate the first row in Table 1, run:
```
python scripts/dispatcher.py -c configs/eval/clip_egnn.json -l ./logs/
```
3. We perform all analysis in the jupyter notebook included [Results.ipynb](analysis/Results.ipynb). We first calculate the hidden representations of the screening using the eval configs above and collect them into one matrix (saved as a pickle file). These are loaded into the jupyter notebook as well as the test set. All tables are then generated in the notebook.
## Downloading Batched AlphaFold Database Structures
Assuming you have a list of uniprot IDs (called `uniprot_ids`) you can run the following to create a .txt file with the Google Storage urls for the AF2 structures:
```
file_paths = [f"gs://public-datasets-deepmind-alphafold-v4/AF-{u}-F1-model_v4.cif" for u in uniprot_ids]
output_file = 'uniprot_cif_paths.txt'
open(output_file, 'w') as file:
file.write('\n'.join(file_paths))
```
Then install gsutils (https://cloud.google.com/storage/docs/gsutil_install) and run the following command:
```
cat uniprot_cif_paths.txt | gsutil -m cp -I /path/to/output/dir/
```
## Citation
```bibtex
@article{mikhael2024clipzyme,
title={CLIPZyme: Reaction-Conditioned Virtual Screening of Enzymes},
author={Mikhael, Peter G and Chinn, Itamar and Barzilay, Regina},
journal={arXiv preprint arXiv:2402.06748},
year={2024}
}
```
Raw data
{
"_id": null,
"home_page": "https://github.com/pgmikhael/clipzyme",
"name": "clipzyme",
"maintainer": null,
"docs_url": null,
"requires_python": "<3.11,>=3.10",
"maintainer_email": null,
"keywords": null,
"author": "Peter G Mikhael",
"author_email": "pgmikhael@csail.mit.edu",
"download_url": "https://files.pythonhosted.org/packages/89/f7/096ec3f2035e75751e70aaf4771dfaf94801a4abfe207f01309bdf5619e5/clipzyme-0.0.11.tar.gz",
"platform": null,
"description": "[](https://github.com/pgmikhael/CLIPZyme/blob/main/LICENSE.txt) \n[](https://arxiv.org/abs/2402.06748)\n<!--  -->\n\n# CLIPZyme\n\nImplementation of the paper [**CLIPZyme: Reaction-Conditioned Virtual Screening of Enzymes**](https://github.com/pgmikhael/CLIPZyme/blob/main/LICENSE.txt)\n\n\n\nTable of contents\n=================\n\n<!--ts-->\n- [CLIPZyme](#clipzyme)\n- [Table of contents](#table-of-contents)\n- [Installation:](#installation)\n- [Checkpoints and Data Files:](#checkpoints-and-data-files)\n- [Screening with CLIPZyme](#screening-with-clipzyme)\n - [Using CLIPZyme's screening set](#using-clipzymes-screening-set)\n - [Using your own screening set](#using-your-own-screening-set)\n - [Interactive (slow)](#interactive-slow)\n - [Batched (fast)](#batched-fast)\n- [Reproducing published results](#reproducing-published-results)\n - [Data processing](#data-processing)\n - [Training and evaluation](#training-and-evaluation)\n - [Downloading Batched AlphaFold Database Structures](#downloading-batched-alphafold-database-structures)\n - [Citation](#citation)\n \n<!--te-->\n\n# Installation:\n\n1. Clone the repository:\n```bash\ngit clone https://github.com/pgmikhael/CLIPZyme.git\n```\n2. Install the dependencies:\n```bash\ncd clipzyme\nconda env create -f environment.yml\nconda activate clipzyme\npython -m pip install clipzyme\n```\n\n3. Download ESM-2 checkpoint `esm2_t33_650M_UR50D`. The `esm_dir` argument should point to this directory. The following command will download the checkpoint directly:\n```bash\nwget https://dl.fbaipublicfiles.com/fair-esm/models/esm2_t33_650M_UR50D.pt\n```\n\n# Checkpoints and Data Files:\n\nThe model checkpoint and data are available on Zenodo [here](https://zenodo.org/records/11187895):\n\n- [clipzyme_data.zip](https://zenodo.org/records/11187895/files/clipzyme_data.zip?download=1):\n - The following commands will download the checkpoint directly: \n ```\n wget https://zenodo.org/records/11187895/files/clipzyme_data.zip\n unzip clipzyme_data.zip -d files\n ```\n - Note that the data files should be extracted into the `files/` directory.\n - `enzymemap.json`: contains the EnzymeMap dataset.\n - `cached_enzymemap.p`: contains the processed EnzymeMap dataset.\n - `clipzyme_screening_set.p`: contains the screening set as dict of UniProt IDs and precomputed protein embeddings.\n - `uniprot2sequence.p`: contains the mapping form sequence ID to amino acids.\n\n- [clipzyme_model.zip](https://zenodo.org/records/11187895/files/clipzyme_model.zip?download=1):\n - The following command will download the checkpoint directly: \n ```\n wget https://zenodo.org/records/11187895/files/clipzyme_model.zip\n unzip clipzyme_model.zip -d files\n ```\n - `clipzyme_model.ckpt`: the trained model checkpoint.\n\n\n\n# Screening with CLIPZyme\n\n## Using CLIPZyme's screening set\n\nFirst, download the screening set and extract the files into `files/`.\n\n\n```python\nimport pickle\nfrom clipzyme import CLIPZyme\n\n## Load the screening set\n##-----------------------\nscreenset = pickle.load(open(\"files/clipzyme_screening_set.p\", 'rb'))\nscreen_hiddens = screenset[\"hiddens\"] # hidden representations (261907, 1280)\nscreen_unis = screenset[\"uniprots\"] # uniprot ids (261907,)\n\n## Load the model and obtain the hidden representations of a reaction\n##-------------------------------------------------------------------\nmodel = CLIPZyme(checkpoint_path=\"files/clipzyme_model.ckpt\")\nreaction = \"[CH3:1][N+:2]([CH3:3])([CH3:4])[CH2:5][CH:6]=[O:7].[O:9]=[O:10].[OH2:8]>>[CH3:1][N+:2]([CH3:3])([CH3:4])[CH2:5][C:6](=[O:7])[OH:8].[OH:9][OH:10]\"\nreaction_embedding = model.extract_reaction_features(reaction=reaction) # (1,1280)\n\nenzyme_scores = screen_hiddens @ reaction_embedding.T # (261907, 1)\n\n```\n\n## Using your own screening set\n\nPrepare your data as a CSV in the following format, and save it as `files/new_data.csv`. For the cases where we wish only to obtain the hidden representations of the sequences, the `reaction` column can be left empty (and vice versa).\n\n| reaction | sequence | protein_id | cif |\n| ------------------------------------------------------------------------------------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------- | -------- |\n| [CH3:1][N+:2]([CH3:3])([CH3:4])[CH2:5][CH:6]=[O:7].[O:9]=[O:10].[OH2:8]>>[CH3:1][N+:2]([CH3:3])([CH3:4])[CH2:5][C:6](=[O:7])[OH:8].[OH:9][OH:10] | MGLSDGEWQLVLNVWGKVEAD<br>IPGHGQEVLIRLFKGHPETLE<br>KFDKFKHLKSEDEMKASEDLK<br>KHGATVLTALGGILKKKGHHE<br>AELKPLAQSHATKHKIPIKYL<br>EFISEAIIHVLHSRHPGDFGA<br>DAQGAMNKALELFRKDIAAKY<br>KELGYQG | P69905 | 1a0s.cif |\n\n\n### Interactive (slow)\n \n```python\nfrom torch.utils.data import DataLoader\nfrom clipzyme import CLIPZyme\nfrom clipzyme import ReactionDataset\nfrom clipzyme.utils.loading import ignore_None_collate\n\n## Create reaction dataset\n#-------------------------\nreaction_dataset = DataLoader(\n ReactionDataset(\n dataset_file_path = \"files/new_data.csv\",\n esm_dir = \"/path/to/esm2_dir\",\n protein_cache_dir = \"/path/to/protein_cache\", # optional, where to cache processed protein graphs\n ),\n batch_size=1,\n collate_fn=ignore_None_collate,\n)\n\n\n## Load the model\n#----------------\nmodel = CLIPZyme(checkpoint_path=\"files/clipzyme_model.ckpt\")\nmodel = model.eval() # optional \n\n## For reaction-enzyme pair\n#--------------------------\nfor batch in reaction_dataset:\n output = model(batch) \n enzyme_scores = output.scores\n protein_hiddens = output.protein_hiddens\n reaction_hiddens = output.reaction_hiddens\n\n## For sequences only\n#--------------------\nfor batch in reaction_dataset:\n protein_hiddens = model.extract_protein_features(batch) \n \n## For reactions only\n#--------------------\nfor batch in reaction_dataset:\n reaction_hiddens = model.extract_reaction_features(batch)\n\n```\n\n### Batched (fast)\n\n1. Update the screening config `configs/screening.json` with the path to your data and indicate what you want to save and where:\n\n\n```JSON\n{\n \"dataset_file_path\": [\"files/new_data.csv\"],\n \"inference_dir\": [\"/where/to/save/embeddings_and_scores\"],\n \"save_hiddens\": [true], # whether to save the hidden representations\n \"save_predictions\": [true], # whether to save the reaction-enzyme pair scores\n \"use_as_protein_encoder\": [true], # whether to use the model as a protein encoder only\n \"use_as_reaction_encoder\": [true], # whether to use the model as a reaction encoder only\n \"esm_dir\": [\"/data/esm/checkpoints\"], path to ESM-2 checkpoints\n \"gpus\": [8], # number of gpus to use,\n \"protein_cache_dir\": [\"/path/to/protein_cache\"], # where to save the protein cache [optional]\n ...\n}\n```\n\nIf you want to use specific GPUs, you can specify them in the `available_gpus` field. For example, to use GPUs 0, 1, and 2, set `available_gpus` to `[\"0,1,2\"]`.\n\n\n\n2. Run the dispatcher with the screening config:\n\n```bash\npython scripts/dispatcher.py -c configs/screening.json -l ./logs/\n```\n\n3. Load the saved embeddings and scores:\n\n```python\nfrom clipzyme import collect_screening_results\n\nscreen_hiddens, screen_unis, enzyme_scores = collect_screening_results(\"configs/screening.json\")\n\n```\n\n\n---------------------\n\n# Reproducing published results\n\n## Data processing\n\nWe obtain the data from the following sources:\n- [EnzymeMap:](`https://doi.org/10.5281/zenodo.7841780`) Heid et al. Enzymemap: Curation, validation and data-driven prediction of enzymatic reactions. 2023.\n- [Terpene Synthases:](`https://zenodo.org/records/10567437`) Samusevich et al. Discovery and characterization of terpene synthases powered by machine learning. 2024. \n\nOur processed data is can be downloaded from [here](https://zenodo.org/records/11187895). \n\n\n## Training and evaluation\n1. To train the models presented in the tables below, run the following command:\n ```\n python scripts/dispatcher.py -c {config_path} -l {log_path}\n ```\n - `{config_path}` is the path to the config file in the table below \n - `{log_path}` is the path in which to save the log file. \n \n For example, to run the first row in Table 1, run:\n ```\n python scripts/dispatcher.py -c configs/train/clip_egnn.json -l ./logs/\n ```\n2. Once you've trained the model, run the eval config to evaluate the model on the test set. For example, to evaluate the first row in Table 1, run:\n ```\n python scripts/dispatcher.py -c configs/eval/clip_egnn.json -l ./logs/\n ```\n3. We perform all analysis in the jupyter notebook included [Results.ipynb](analysis/Results.ipynb). We first calculate the hidden representations of the screening using the eval configs above and collect them into one matrix (saved as a pickle file). These are loaded into the jupyter notebook as well as the test set. All tables are then generated in the notebook.\n\n## Downloading Batched AlphaFold Database Structures\n\nAssuming you have\u00a0a list of uniprot IDs\u00a0(called `uniprot_ids`) you can run the following to create a .txt file with the Google Storage urls for the AF2 structures:\n```\nfile_paths = [f\"gs://public-datasets-deepmind-alphafold-v4/AF-{u}-F1-model_v4.cif\" for u in uniprot_ids]\noutput_file = 'uniprot_cif_paths.txt' \nopen(output_file, 'w') as file:\n file.write('\\n'.join(file_paths))\n```\nThen install gsutils\u00a0(https://cloud.google.com/storage/docs/gsutil_install) and run the following command:\n```\ncat\u00a0uniprot_cif_paths.txt | gsutil -m cp -I\u00a0/path/to/output/dir/\n```\n\n\n## Citation\n\n```bibtex\n@article{mikhael2024clipzyme,\n title={CLIPZyme: Reaction-Conditioned Virtual Screening of Enzymes},\n author={Mikhael, Peter G and Chinn, Itamar and Barzilay, Regina},\n journal={arXiv preprint arXiv:2402.06748},\n year={2024}\n}\n```\n",
"bugtrack_url": null,
"license": "Apache 2.0",
"summary": "Reaction-Conditioned Virtual Screening of Enzymes",
"version": "0.0.11",
"project_urls": {
"Homepage": "https://github.com/pgmikhael/clipzyme",
"Repository": "https://github.com/pgmikhael/clipzyme"
},
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "a3f578942d73c6b7573c977d90ce30df17ace6f8001bbbe3fd4c7f12ead94457",
"md5": "f45572683817c0f1d97014c228629f79",
"sha256": "c433820e54c419c35e1094504d41fc986e5522eb59a9d6d0de3a94527b77db31"
},
"downloads": -1,
"filename": "clipzyme-0.0.11-py3-none-any.whl",
"has_sig": false,
"md5_digest": "f45572683817c0f1d97014c228629f79",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": "<3.11,>=3.10",
"size": 128053,
"upload_time": "2024-08-01T05:47:11",
"upload_time_iso_8601": "2024-08-01T05:47:11.898507Z",
"url": "https://files.pythonhosted.org/packages/a3/f5/78942d73c6b7573c977d90ce30df17ace6f8001bbbe3fd4c7f12ead94457/clipzyme-0.0.11-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "89f7096ec3f2035e75751e70aaf4771dfaf94801a4abfe207f01309bdf5619e5",
"md5": "c2e71ed15fe3a06a48993c6a90a28e40",
"sha256": "e3469be34f78fcb60875b3bffe7090b6d9fbd886eba7b7d76378d0d1b6194eee"
},
"downloads": -1,
"filename": "clipzyme-0.0.11.tar.gz",
"has_sig": false,
"md5_digest": "c2e71ed15fe3a06a48993c6a90a28e40",
"packagetype": "sdist",
"python_version": "source",
"requires_python": "<3.11,>=3.10",
"size": 108278,
"upload_time": "2024-08-01T05:47:13",
"upload_time_iso_8601": "2024-08-01T05:47:13.492586Z",
"url": "https://files.pythonhosted.org/packages/89/f7/096ec3f2035e75751e70aaf4771dfaf94801a4abfe207f01309bdf5619e5/clipzyme-0.0.11.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-08-01 05:47:13",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "pgmikhael",
"github_project": "clipzyme",
"travis_ci": false,
"coveralls": false,
"github_actions": false,
"lcname": "clipzyme"
}
JSON
