# Using DeepMol models
Models available so far:
| Model Name | How to Call | Prediction Type |
|---------------------------------------------|---------------------------------|----------------------------------------------------------------|
| BBB (Blood-Brain Barrier) | `BBB` | Penetrates BBB (1) or does not penetrate BBB (0) |
| AMES Mutagenicity | `AMES` | Mutagenic (1) or not mutagenic (0) |
| Human plasma protein binding rate (PPBR) | `PPBR` | Rate of PPBR expressed in percentage |
| Volume of Distribution (VD) at steady state | `VDss` | Volume of Distribution expressed in liters per kilogram (L/kg)|
| Caco-2 (Cell Effective Permeability) | `Caco2` | Cell Effective Permeability (cm/s) |
| HIA (Human Intestinal Absorption) | `HIA` | Absorbed (1) or not absorbed (0) |
| Bioavailability | `Bioavailability` | Bioavailable (1) or not bioavailable (0) |
| Lipophilicity | `Lipophilicity` | Lipophilicity log-ratio |
| Solubility | `Solubility` | Solubility (log mol/L) |
| CYP P450 2C9 Inhibition | `CYP2C9Inhibition` | Inhibit (1) or does not inhibit (0) |
| CYP P450 3A4 Inhibition | `CYP3A4Inhibition` | Inhibit (1) or does not inhibit (0) |
| CYP2C9 Substrate | `CYP2C9Substrate`| Metabolized (1) or does not metabolize (0) |
| CYP2D6 Substrate | `CYP2D6Substrate`| Metabolized (1) or does not metabolize (0) |
| CYP3A4 Substrate | `CYP3A4Substrate`| Metabolized (1) or does not metabolize (0) |
| Hepatocyte Clearance | `HepatocyteClearance` | Drug hepatocyte clearance (uL.min-1.(10^6 cells)-1) |
| NPClassifier | `NPClassifier` | Pathway, Superclass, Class |
| Plants secondary metabolite precursors predictor | `PlantsSMPrecursorPredictor` | Precursor 1; Precursor 2 |
| Microsome Clearance | `MicrosomeClearance` | Drug microsome clearance (mL.min-1.g-1) |
| LD50 | `LD50` | LD50 (log(1/(mol/kg))) |
| hERG Blockers | `hERGBlockers` | hERG blocker (1) or not blocker (0) |
## How to use:
You can use them either individually or mixed together.
You can call one model individually, pass a CSV file and get the results in one dataframe:
```python
from deepmol_models import BBB
results = BBB().predict_from_csv("dataset.csv", smiles_field="Drug", id_field="Drug_ID", output_file="predictions.csv")
results
```
| ID | SMILES | BBB Penetration |
|----|-------------------------------------------------|-----------------|
| 0 | OCC(S)CS | 1.0 |
| 1 | CC[N+](C)(C)c1cccc(O)c1 | 0.0 |
| 2 | Nc1ncnc2c1ncn2[C@@H]1O[C@H](CO)[C@@H](O)[C@@H]1O | 1.0 |
| 3 | CC(=O)OCC1=C(C(=O)O)N2C(=O)[C@@H](NC(=O)CC#N)[...| 0.0 |
| 4 | CC1(C)S[C@@H]2[C@H](NC(=O)[C@H](C(=O)O)c3ccsc3...| 0.0 |
Or pass SMILES strings and get the results in one dataframe:
```python
from deepmol_models import BBB
results = BBB().predict_from_csv("dataset.csv", smiles_field="Drug", id_field="Drug_ID", output_file="predictions.csv")
results
```
| ID | SMILES | BBB Penetration |
|----|-------------------------------------------------|-----------------|
| 1 | CC[N+](C)(C)c1cccc(O)c1 | 0.0 |
| 2 | Nc1ncnc2c1ncn2[C@@H]1O[C@H](CO)[C@@H](O)[C@@H]1O | 1.0 |
Complementarily, you can run several models:
```python
from deepmol_models import BBB, PPBR, VDss, Caco2, HIA, Bioavailability, \
Lipophilicity, Solubility, PlantsSMPrecursorPredictor, NPClassifier, MixedPredictor
# results = MixedPredictor([BBB(), Caco2(), CYP2D6Inhibition(), NPClassifier()]).predict_from_csv("test_molecules.csv", "Drug", "Drug_ID", output_file="predictions.csv")
results = MixedPredictor([BBB(), PPBR(), VDss(), Caco2(),
HIA(), Bioavailability(), Lipophilicity(),
Solubility(), PlantsSMPrecursorPredictor(), NPClassifier()]).predict_from_csv("test_molecules.csv", smiles_field="Drug", id_field="Drug_ID", output_file="predictions.csv")
results
```
| ID | SMILES | BBB Penetration | Human PPBR | VDss | Cell Effective Permeability | Human Intestinal Absorption | Bioavailability | Lipophilicity | Solubility | Precursors | Pathways | Superclass | Class |
|----|-------------------------------------------------|-----------------|------------|----------|-----------------------------|-----------------------------|-----------------|---------------|------------|---------------------------------------|--------------------------|-----------------------|-----------------------|
| 0 | OCC(S)CS | 1.0 | 64.832665 | 5.803529 | -4.725497 | 0.0 | 0.0 | 0.290602 | 0.117866 | | Fatty acyls | Fatty alcohols | Fatty alcohols |
| 1 | CC[N+](C)(C)c1cccc(O)c1 | 0.0 | 32.882912 | 2.891243 | -4.989814 | 0.0 | 0.0 | 0.271549 | -0.606772 | L-Lysine | Alkaloids | Tyrosine alkaloids | Phenylethylamines |
| 2 | Nc1ncnc2c1ncn2[C@@H]1O[C@H](CO)[C@@H](O)[C@@H]1O | 1.0 | 41.540812 | 2.722000 | -6.059630 | 1.0 | 0.0 | 0.250526 | -1.701435 | Dimethylallyl diphosphate | Carbohydrates | Nucleosides | Purine nucleosides |
| 3 | CC(=O)OCC1=C(C(=O)O)N2C(=O)[C@@H](NC(=O)CC#N)[...] | 0.0 | 52.921825 | 0.329071 | -5.473660 | 0.0 | 0.0 | 0.267842 | -2.512460 | Geranylgeranyl diphosphate; L-Alanine | Amino acids and Peptides | β-lactams | Cephalosporins |
## Visualization
You can use our API to access the bokeh representation of the chemical space and check some features of the molecules:
```python
from deepmol_models import bokeh_plot
bokeh_plot(results, "Solubility", additional_labels=["Pathways", "Superclass", "Class"])
```
Raw data
{
"_id": null,
"home_page": null,
"name": "deepmol-models",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": null,
"keywords": "machine-learning, deep-learning, cheminformatics, drug-discovery",
"author": "DeepMol Team",
"author_email": null,
"download_url": "https://files.pythonhosted.org/packages/de/99/93de4992845f75117ac0a72703f6aec031ed7ce52ee3e776df02eaf1f777/deepmol_models-0.0.5.tar.gz",
"platform": "unix",
"description": "# Using DeepMol models\n\nModels available so far: \n\n| Model Name | How to Call | Prediction Type |\n|---------------------------------------------|---------------------------------|----------------------------------------------------------------|\n| BBB (Blood-Brain Barrier) | `BBB` | Penetrates BBB (1) or does not penetrate BBB (0) |\n| AMES Mutagenicity | `AMES` | Mutagenic (1) or not mutagenic (0) |\n| Human plasma protein binding rate (PPBR) | `PPBR` | Rate of PPBR expressed in percentage |\n| Volume of Distribution (VD) at steady state | `VDss` | Volume of Distribution expressed in liters per kilogram (L/kg)|\n| Caco-2 (Cell Effective Permeability) | `Caco2` | Cell Effective Permeability (cm/s) |\n| HIA (Human Intestinal Absorption) | `HIA` | Absorbed (1) or not absorbed (0) |\n| Bioavailability | `Bioavailability` | Bioavailable (1) or not bioavailable (0) |\n| Lipophilicity | `Lipophilicity` | Lipophilicity log-ratio |\n| Solubility | `Solubility` | Solubility (log mol/L) |\n| CYP P450 2C9 Inhibition | `CYP2C9Inhibition` | Inhibit (1) or does not inhibit (0) |\n| CYP P450 3A4 Inhibition | `CYP3A4Inhibition` | Inhibit (1) or does not inhibit (0) |\n| CYP2C9 Substrate | `CYP2C9Substrate`| Metabolized (1) or does not metabolize (0) |\n| CYP2D6 Substrate | `CYP2D6Substrate`| Metabolized (1) or does not metabolize (0) |\n| CYP3A4 Substrate | `CYP3A4Substrate`| Metabolized (1) or does not metabolize (0) |\n| Hepatocyte Clearance | `HepatocyteClearance` | Drug hepatocyte clearance (uL.min-1.(10^6 cells)-1) |\n| NPClassifier | `NPClassifier` | Pathway, Superclass, Class |\n| Plants secondary metabolite precursors predictor | `PlantsSMPrecursorPredictor` | Precursor 1; Precursor 2 |\n| Microsome Clearance | `MicrosomeClearance` | Drug microsome clearance (mL.min-1.g-1) |\n| LD50 | `LD50` | LD50 (log(1/(mol/kg))) |\n| hERG Blockers | `hERGBlockers` | hERG blocker (1) or not blocker (0) |\n\n## How to use:\n\nYou can use them either individually or mixed together. \n\nYou can call one model individually, pass a CSV file and get the results in one dataframe:\n\n```python\nfrom deepmol_models import BBB\nresults = BBB().predict_from_csv(\"dataset.csv\", smiles_field=\"Drug\", id_field=\"Drug_ID\", output_file=\"predictions.csv\")\nresults\n```\n\n| ID | SMILES | BBB Penetration |\n|----|-------------------------------------------------|-----------------|\n| 0 | OCC(S)CS | 1.0 |\n| 1 | CC[N+](C)(C)c1cccc(O)c1 | 0.0 |\n| 2 | Nc1ncnc2c1ncn2[C@@H]1O[C@H](CO)[C@@H](O)[C@@H]1O | 1.0 |\n| 3 | CC(=O)OCC1=C(C(=O)O)N2C(=O)[C@@H](NC(=O)CC#N)[...| 0.0 |\n| 4 | CC1(C)S[C@@H]2[C@H](NC(=O)[C@H](C(=O)O)c3ccsc3...| 0.0 |\n\n\nOr pass SMILES strings and get the results in one dataframe:\n\n```python\nfrom deepmol_models import BBB\nresults = BBB().predict_from_csv(\"dataset.csv\", smiles_field=\"Drug\", id_field=\"Drug_ID\", output_file=\"predictions.csv\")\nresults\n```\n\n| ID | SMILES | BBB Penetration |\n|----|-------------------------------------------------|-----------------|\n| 1 | CC[N+](C)(C)c1cccc(O)c1 | 0.0 |\n| 2 | Nc1ncnc2c1ncn2[C@@H]1O[C@H](CO)[C@@H](O)[C@@H]1O | 1.0 |\n\nComplementarily, you can run several models:\n\n```python\nfrom deepmol_models import BBB, PPBR, VDss, Caco2, HIA, Bioavailability, \\\n Lipophilicity, Solubility, PlantsSMPrecursorPredictor, NPClassifier, MixedPredictor\n\n# results = MixedPredictor([BBB(), Caco2(), CYP2D6Inhibition(), NPClassifier()]).predict_from_csv(\"test_molecules.csv\", \"Drug\", \"Drug_ID\", output_file=\"predictions.csv\")\nresults = MixedPredictor([BBB(), PPBR(), VDss(), Caco2(), \n HIA(), Bioavailability(), Lipophilicity(),\n Solubility(), PlantsSMPrecursorPredictor(), NPClassifier()]).predict_from_csv(\"test_molecules.csv\", smiles_field=\"Drug\", id_field=\"Drug_ID\", output_file=\"predictions.csv\")\nresults\n```\n\n| ID | SMILES | BBB Penetration | Human PPBR | VDss | Cell Effective Permeability | Human Intestinal Absorption | Bioavailability | Lipophilicity | Solubility | Precursors | Pathways | Superclass | Class |\n|----|-------------------------------------------------|-----------------|------------|----------|-----------------------------|-----------------------------|-----------------|---------------|------------|---------------------------------------|--------------------------|-----------------------|-----------------------|\n| 0 | OCC(S)CS | 1.0 | 64.832665 | 5.803529 | -4.725497 | 0.0 | 0.0 | 0.290602 | 0.117866 | | Fatty acyls | Fatty alcohols | Fatty alcohols |\n| 1 | CC[N+](C)(C)c1cccc(O)c1 | 0.0 | 32.882912 | 2.891243 | -4.989814 | 0.0 | 0.0 | 0.271549 | -0.606772 | L-Lysine | Alkaloids | Tyrosine alkaloids | Phenylethylamines |\n| 2 | Nc1ncnc2c1ncn2[C@@H]1O[C@H](CO)[C@@H](O)[C@@H]1O | 1.0 | 41.540812 | 2.722000 | -6.059630 | 1.0 | 0.0 | 0.250526 | -1.701435 | Dimethylallyl diphosphate | Carbohydrates | Nucleosides | Purine nucleosides |\n| 3 | CC(=O)OCC1=C(C(=O)O)N2C(=O)[C@@H](NC(=O)CC#N)[...] | 0.0 | 52.921825 | 0.329071 | -5.473660 | 0.0 | 0.0 | 0.267842 | -2.512460 | Geranylgeranyl diphosphate; L-Alanine | Amino acids and Peptides | \u03b2-lactams | Cephalosporins |\n\n\n## Visualization\n\nYou can use our API to access the bokeh representation of the chemical space and check some features of the molecules:\n\n```python\nfrom deepmol_models import bokeh_plot\nbokeh_plot(results, \"Solubility\", additional_labels=[\"Pathways\", \"Superclass\", \"Class\"])\n```\n\n\n",
"bugtrack_url": null,
"license": "BSD 2-Clause License",
"summary": "DeepMol Case Studies",
"version": "0.0.5",
"project_urls": null,
"split_keywords": [
"machine-learning",
" deep-learning",
" cheminformatics",
" drug-discovery"
],
"urls": [
{
"comment_text": null,
"digests": {
"blake2b_256": "5c14fced0262a5416e2849223ddf1babbdf9b77c01291b0ea690d05b8c926cbf",
"md5": "104e66dd7946453d4e0f161fa27fa459",
"sha256": "37885f8592bedcc55a156222d404b7d2e55c74462dd1829891fb9929ca973aef"
},
"downloads": -1,
"filename": "deepmol_models-0.0.5-py3-none-any.whl",
"has_sig": false,
"md5_digest": "104e66dd7946453d4e0f161fa27fa459",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 31388,
"upload_time": "2025-01-23T19:58:41",
"upload_time_iso_8601": "2025-01-23T19:58:41.208821Z",
"url": "https://files.pythonhosted.org/packages/5c/14/fced0262a5416e2849223ddf1babbdf9b77c01291b0ea690d05b8c926cbf/deepmol_models-0.0.5-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "de9993de4992845f75117ac0a72703f6aec031ed7ce52ee3e776df02eaf1f777",
"md5": "aa0a5739ffcf0d7475b3fb6c721ba608",
"sha256": "e2e0137c3c018ef9bc10360a16f631eeea0ae2b2165cdcc48f32b379e42bd294"
},
"downloads": -1,
"filename": "deepmol_models-0.0.5.tar.gz",
"has_sig": false,
"md5_digest": "aa0a5739ffcf0d7475b3fb6c721ba608",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 30398,
"upload_time": "2025-01-23T19:58:43",
"upload_time_iso_8601": "2025-01-23T19:58:43.155357Z",
"url": "https://files.pythonhosted.org/packages/de/99/93de4992845f75117ac0a72703f6aec031ed7ce52ee3e776df02eaf1f777/deepmol_models-0.0.5.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-01-23 19:58:43",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "deepmol-models"
}
JSON