unimol

Name: unimol
Author: jinzhezenggroup

113stars23forks

How to install unimol?

jinzhezenggroup/computational-chemistry-agent-skills/unimol

$npx -y skills add jinzhezenggroup/computational-chemistry-agent-skills --skill unimol

Installs into the current project.

Summary

A standardized CLI wrapper for Uni-Mol molecular ML workflows that handles representation extraction (embeddings), model training (regression/classification), and property prediction with built-in RDKit SMILES validation. USE WHEN you need to generate molecular embeddings, train

Files · 1

View on GitHub

SKILL.md

1	# Uni-Mol
2
3	This skill provides practical command patterns for Uni-Mol molecular representation / training / prediction using the standardized CLI wrapper: `<skill_path>/scripts/unimol_helper.py`.
4
5	Key behaviors (important for Agents):
6
7	- The script prints environment detection (Python/Torch/CUDA) by default.
8	- Bad/illegal SMILES are skipped and logged to `*.skipped.csv` (no crash).
9	- Each run ends by printing absolute output paths like:
10	- `[RESULT] repr_npy=/abs/path.npy`
11	- `[RESULT] model_dir=/abs/model_dir`
12	- `[RESULT] pred_csv=/abs/pred.csv`
13
14	## Quick Start
15
16	Check CLI help:
17
18	```bash
19	uv run python <skill_path>/scripts/unimol_helper.py --help
20	```
21
22	Check subcommand help:
23
24	```bash
25	uv run python <skill_path>/scripts/unimol_helper.py repr --help
26	uv run python <skill_path>/scripts/unimol_helper.py train --help
27	uv run python <skill_path>/scripts/unimol_helper.py predict --help
28	```
29
30	Disable environment printing (optional):
31
32	```bash
33	uv run python <skill_path>/scripts/unimol_helper.py --no-env repr --smiles "CCO" --output out.npy
34	```
35
36	## Core Tasks
37
38	### 1) Extract molecular representations (embedding) to .npy
39
40	Single SMILES:
41
42	```bash
43	uv run python <skill_path>/scripts/unimol_helper.py repr \
44	--smiles "CCO" \
45	--output /tmp/ccO.repr.npy
46	```
47
48	From CSV (default SMILES column is `smiles`):
49
50	```bash
51	uv run python <skill_path>/scripts/unimol_helper.py repr \
52	--file data.csv \
53	--smiles-col smiles \
54	--output data.repr.npy
55	```
56
57	From SMI:
58
59	```bash
60	uv run python <skill_path>/scripts/unimol_helper.py repr \
61	--file molecules.smi \
62	--output molecules.repr.npy
63	```
64
65	Force CPU / GPU:
66
67	```bash
68	# Force CPU
69	uv run python <skill_path>/scripts/unimol_helper.py repr --smiles "CCO" --no-gpu --output out.npy
70
71	# Force GPU (will warn & fall back if CUDA is unavailable)
72	uv run python <skill_path>/scripts/unimol_helper.py repr --smiles "CCO" --use-gpu --output out.npy
73	```
74
75	### 2) Train a property model (classification / regression / multilabel\_\*)
76
77	Regression training (CSV must contain `smiles` and `target` columns):
78
79	```bash
80	uv run python <skill_path>/scripts/unimol_helper.py train \
81	--task regression \
82	--input train.csv \
83	--smiles-col smiles \
84	--target-col target \
85	--epochs 50 \
86	--output ./model_reg
87	```
88
89	Classification training:
90
91	```bash
92	uv run python <skill_path>/scripts/unimol_helper.py train \
93	--task classification \
94	--input train.csv \
95	--smiles-col smiles \
96	--target-col target \
97	--epochs 50 \
98	--output ./model_cls
99	```
100
101	Multilabel regression training (explicit multi-target columns):
102
103	```bash
104	uv run python <skill_path>/scripts/unimol_helper.py train \
105	--task multilabel_regression \
106	--input train.csv \
107	--smiles-col smiles \
108	--target-cols target_0,target_1,target_2 \
109	--epochs 50 \
110	--output ./model_mreg
111	```
112
113	Multilabel classification training:
114
115	```bash
116	uv run python <skill_path>/scripts/unimol_helper.py train \
117	--task multilabel_classification \
118	--input train.csv \
119	--smiles-col smiles \
120	--target-cols y_cls_0,y_cls_1,y_cls_2 \
121	--epochs 50 \
122	--output ./model_mcls
123	```
124
125	Target recognition for training:
126
127	- Single-task (`classification` / `regression`): use `--target-col` (default `target`).
128	- Multilabel tasks: prefer `--target-cols` (comma-separated).
129	- If `--target-cols` is omitted for multilabel tasks, the helper auto-detects columns named `target` or prefixed with `target_` (case-insensitive).
130
131	Force CPU:
132
133	```bash
134	uv run python <skill_path>/scripts/unimol_helper.py train \
135	--task regression \
136	--input train.csv \
137	--epochs 50 \
138	--output ./model_cpu \
139	--no-cuda
140	```
141
142	### 3) Predict properties to .csv
143
144	Predict from CSV:
145
146	```bash
147	uv run python <skill_path>/scripts/unimol_helper.py predict \
148	--model ./model_reg \
149	--input test.csv \
150	--smiles-col smiles \
151	--output pred.csv
152	```
153
154	Predict from SMI:
155
156	```bash
157	uv run python <skill_path>/scripts/unimol_helper.py predict \
158	--model ./model_reg \
159	--input test.smi \
160	--output pred.csv
161	```
162
163	Notes:
164
165	- Output CSV contains the input rows (for valid SMILES) plus `pred` / `pred_*` columns.
166	- If there are bad SMILES, they are skipped and saved to `pred.csv.skipped.csv` (or your `--error-log` path).
167
168	## Agent Checklist
169
170	When usi

People who liked this also viewed

6 picks

Type

unimol

113stars23forks

View on GitHub

How to install unimol?

jinzhezenggroup/computational-chemistry-agent-skills/unimol

$npx -y skills add jinzhezenggroup/computational-chemistry-agent-skills --skill unimol

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Uni-Mol
2
3	This skill provides practical command patterns for Uni-Mol molecular representation / training / prediction using the standardized CLI wrapper: `<skill_path>/scripts/unimol_helper.py`.
4
5	Key behaviors (important for Agents):
6
7	- The script prints environment detection (Python/Torch/CUDA) by default.
8	- Bad/illegal SMILES are skipped and logged to `*.skipped.csv` (no crash).
9	- Each run ends by printing absolute output paths like:
10	- `[RESULT] repr_npy=/abs/path.npy`
11	- `[RESULT] model_dir=/abs/model_dir`
12	- `[RESULT] pred_csv=/abs/pred.csv`
13
14	## Quick Start
15
16	Check CLI help:
17
18	```bash
19	uv run python <skill_path>/scripts/unimol_helper.py --help
20	```
21
22	Check subcommand help:
23
24	```bash
25	uv run python <skill_path>/scripts/unimol_helper.py repr --help
26	uv run python <skill_path>/scripts/unimol_helper.py train --help
27	uv run python <skill_path>/scripts/unimol_helper.py predict --help
28	```
29
30	Disable environment printing (optional):
31
32	```bash
33	uv run python <skill_path>/scripts/unimol_helper.py --no-env repr --smiles "CCO" --output out.npy
34	```
35
36	## Core Tasks
37
38	### 1) Extract molecular representations (embedding) to .npy
39
40	Single SMILES:
41
42	```bash
43	uv run python <skill_path>/scripts/unimol_helper.py repr \
44	--smiles "CCO" \
45	--output /tmp/ccO.repr.npy
46	```
47
48	From CSV (default SMILES column is `smiles`):
49
50	```bash
51	uv run python <skill_path>/scripts/unimol_helper.py repr \
52	--file data.csv \
53	--smiles-col smiles \
54	--output data.repr.npy
55	```
56
57	From SMI:
58
59	```bash
60	uv run python <skill_path>/scripts/unimol_helper.py repr \
61	--file molecules.smi \
62	--output molecules.repr.npy
63	```
64
65	Force CPU / GPU:
66
67	```bash
68	# Force CPU
69	uv run python <skill_path>/scripts/unimol_helper.py repr --smiles "CCO" --no-gpu --output out.npy
70
71	# Force GPU (will warn & fall back if CUDA is unavailable)
72	uv run python <skill_path>/scripts/unimol_helper.py repr --smiles "CCO" --use-gpu --output out.npy
73	```
74
75	### 2) Train a property model (classification / regression / multilabel\_\*)
76
77	Regression training (CSV must contain `smiles` and `target` columns):
78
79	```bash
80	uv run python <skill_path>/scripts/unimol_helper.py train \
81	--task regression \
82	--input train.csv \
83	--smiles-col smiles \
84	--target-col target \
85	--epochs 50 \
86	--output ./model_reg
87	```
88
89	Classification training:
90
91	```bash
92	uv run python <skill_path>/scripts/unimol_helper.py train \
93	--task classification \
94	--input train.csv \
95	--smiles-col smiles \
96	--target-col target \
97	--epochs 50 \
98	--output ./model_cls
99	```
100
101	Multilabel regression training (explicit multi-target columns):
102
103	```bash
104	uv run python <skill_path>/scripts/unimol_helper.py train \
105	--task multilabel_regression \
106	--input train.csv \
107	--smiles-col smiles \
108	--target-cols target_0,target_1,target_2 \
109	--epochs 50 \
110	--output ./model_mreg
111	```
112
113	Multilabel classification training:
114
115	```bash
116	uv run python <skill_path>/scripts/unimol_helper.py train \
117	--task multilabel_classification \
118	--input train.csv \
119	--smiles-col smiles \
120	--target-cols y_cls_0,y_cls_1,y_cls_2 \
121	--epochs 50 \
122	--output ./model_mcls
123	```
124
125	Target recognition for training:
126
127	- Single-task (`classification` / `regression`): use `--target-col` (default `target`).
128	- Multilabel tasks: prefer `--target-cols` (comma-separated).
129	- If `--target-cols` is omitted for multilabel tasks, the helper auto-detects columns named `target` or prefixed with `target_` (case-insensitive).
130
131	Force CPU:
132
133	```bash
134	uv run python <skill_path>/scripts/unimol_helper.py train \
135	--task regression \
136	--input train.csv \
137	--epochs 50 \
138	--output ./model_cpu \
139	--no-cuda
140	```
141
142	### 3) Predict properties to .csv
143
144	Predict from CSV:
145
146	```bash
147	uv run python <skill_path>/scripts/unimol_helper.py predict \
148	--model ./model_reg \
149	--input test.csv \
150	--smiles-col smiles \
151	--output pred.csv
152	```
153
154	Predict from SMI:
155
156	```bash
157	uv run python <skill_path>/scripts/unimol_helper.py predict \
158	--model ./model_reg \
159	--input test.smi \
160	--output pred.csv
161	```
162
163	Notes:
164
165	- Output CSV contains the input rows (for valid SMILES) plus `pred` / `pred_*` columns.
166	- If there are bad SMILES, they are skipped and saved to `pred.csv.skipped.csv` (or your `--error-log` path).
167
168	## Agent Checklist
169
170	When usi

People who liked this also viewed

6 picks

Type