deepchem-molgraph-featurization

Name: deepchem-molgraph-featurization
Author: ma-compbio-lab

36stars5forks

How to install deepchem-molgraph-featurization?

ma-compbio-lab/skillfoundry/deepchem-molgraph-featurization

$npx -y skills add ma-compbio-lab/SkillFoundry --skill deepchem-molgraph-featurization

Installs into the current project.

Summary

Use this skill to featurize SMILES strings into DeepChem molecular graph objects. Prefer it for local graph-based preprocessing before molecular machine-learning experiments.

Files · 1

View on GitHub

SKILL.md

1	## Purpose
2	Create deterministic DeepChem molecular graph features from a small SMILES table using the repo-managed `chemtools` prefix.
3
4	## When to use
5	- You need a lightweight DeepChem starter without training a model.
6	- You want to inspect graph sizes and node-feature dimensions before modeling.
7
8	## When not to use
9	- You need trained DeepChem models requiring TensorFlow or PyTorch.
10	- You need dataset download or benchmark automation.
11
12	## Inputs
13	- A TSV file with `molecule_id` and `smiles` columns.
14
15	## Outputs
16	- A JSON summary with graph dimensions for each molecule.
17
18	## Requirements
19	- `slurm/envs/chemtools`
20	- DeepChem and RDKit in that prefix
21
22	## Procedure
23	1. Run `slurm/envs/chemtools/bin/python skills/computational-chemistry-and-molecular-simulation/deepchem-molgraph-featurization/scripts/featurize_molecules.py --input skills/computational-chemistry-and-molecular-simulation/deepchem-molgraph-featurization/examples/molecules.tsv --out scratch/deepchem/featurization.json`.
24	2. Inspect node counts, edge counts, and feature dimensions.
25	3. Use the feature summary as a preflight step before larger molecular ML runs.
26
27	## Validation
28	- The script exits successfully.
29	- Each molecule yields graph metadata.
30	- Node and edge counts are positive for valid molecules.
31
32	## Failure modes and fixes
33	- Missing optional ML backends: this skill only requires the featurizer path, not `torch` or `tensorflow`.
34	- Invalid SMILES: correct the input row before featurization.
35
36	## Provenance
37	- DeepChem documentation: https://deepchem.readthedocs.io/en/latest/
38
39	## Related skills
40	- `rdkit-molecular-descriptors`

People who liked this also viewed

6 picks

Type

deepchem-molgraph-featurization

36stars5forks

View on GitHub

How to install deepchem-molgraph-featurization?

ma-compbio-lab/skillfoundry/deepchem-molgraph-featurization

$npx -y skills add ma-compbio-lab/SkillFoundry --skill deepchem-molgraph-featurization

Installs into the current project.

Summary

Use this skill to featurize SMILES strings into DeepChem molecular graph objects. Prefer it for local graph-based preprocessing before molecular machine-learning experiments.

Files · 1

View on GitHub

SKILL.md

1	## Purpose
2	Create deterministic DeepChem molecular graph features from a small SMILES table using the repo-managed `chemtools` prefix.
3
4	## When to use
5	- You need a lightweight DeepChem starter without training a model.
6	- You want to inspect graph sizes and node-feature dimensions before modeling.
7
8	## When not to use
9	- You need trained DeepChem models requiring TensorFlow or PyTorch.
10	- You need dataset download or benchmark automation.
11
12	## Inputs
13	- A TSV file with `molecule_id` and `smiles` columns.
14
15	## Outputs
16	- A JSON summary with graph dimensions for each molecule.
17
18	## Requirements
19	- `slurm/envs/chemtools`
20	- DeepChem and RDKit in that prefix
21
22	## Procedure
23	1. Run `slurm/envs/chemtools/bin/python skills/computational-chemistry-and-molecular-simulation/deepchem-molgraph-featurization/scripts/featurize_molecules.py --input skills/computational-chemistry-and-molecular-simulation/deepchem-molgraph-featurization/examples/molecules.tsv --out scratch/deepchem/featurization.json`.
24	2. Inspect node counts, edge counts, and feature dimensions.
25	3. Use the feature summary as a preflight step before larger molecular ML runs.
26
27	## Validation
28	- The script exits successfully.
29	- Each molecule yields graph metadata.
30	- Node and edge counts are positive for valid molecules.
31
32	## Failure modes and fixes
33	- Missing optional ML backends: this skill only requires the featurizer path, not `torch` or `tensorflow`.
34	- Invalid SMILES: correct the input row before featurization.
35
36	## Provenance
37	- DeepChem documentation: https://deepchem.readthedocs.io/en/latest/
38
39	## Related skills
40	- `rdkit-molecular-descriptors`

People who liked this also viewed

6 picks

Type