molecular-descriptors

Name: molecular-descriptors
Author: GPTomics

1.1kstars185forks

How to install molecular-descriptors?

gptomics/bioskills/molecular-descriptors

$npx -y skills add GPTomics/bioSkills --skill molecular-descriptors

Installs into the current project.

Summary

Calculates molecular fingerprints (ECFP/Morgan, FCFP, MACCS, RDKit, AtomPair, TopologicalTorsion, Avalon, MAP4, MHFP6) and physicochemical descriptors (Lipinski, QED, TPSA, Crippen LogP, 3D shape) with explicit choice tables, bit vs count semantics, and partial-charge model selec

Files · 1

View on GitHub

SKILL.md

1	## Version Compatibility
2
3	Reference examples tested with: RDKit 2024.09+, numpy 1.26+, pandas 2.2+, map4 1.1+ (MAP4), mhfp 1.9+. Use `mapchiral` separately when the stereochemistry-aware MAP4C fingerprint is intended.
4
5	Before using code patterns, verify installed versions match. If versions differ:
6	- Python: `pip show <package>` then `help(module.function)` to check signatures
7
8	If code throws ImportError, AttributeError, or TypeError, introspect the installed
9	package and adapt the example to match the actual API rather than retrying.
10
11	# Molecular Descriptors
12
13	Featurize molecules for similarity search, QSAR, virtual screening, or ML. Fingerprint performance is dataset- and objective-dependent: ECFP4 is a strong drug-like baseline, atom-pair and topological-torsion fingerprints expose longer-range topology, MAP4/MHFP6 target broader chemical-space searches, and 3D conformer-based descriptors are needed when shape and stereochemistry matter.
14
15	For canonicalization before featurization, see `chemoinformatics/molecular-standardization`. For 3D-only descriptors, see `chemoinformatics/conformer-generation`.
16
17	## Fingerprint Taxonomy
18
19	\| Fingerprint \| Type \| Radius/Path \| Bits \| Use case \| Fails when \|
20	\|-------------\|------\|-------------\|------\|----------\|------------\|
21	\| Morgan (ECFP) \| Circular \| r=2 (ECFP4), r=3 (ECFP6) \| 2048 typical \| Drug-like similarity, ML default \| Loses long-range topology; bit collisions at low nBits \|
22	\| FCFP \| Functional Morgan \| r=2 default \| 2048 \| Pharmacophore-aware similarity \| Same caveats as ECFP; less specific \|
23	\| MACCS \| Substructure key \| 166 fixed bits \| 167 \| Quick fingerprint, drug-likeness \| Too sparse for large diverse libraries \|
24	\| RDKit FP \| Path/subgraph-based \| paths and branched subgraphs up to 7 bonds by default \| 2048 \| RDKit-native ECFP alternative \| Drug-like only; not optimal for scaffold hopping \|
25	\| AtomPair \| Pair + topological distance \| All atom pairs \| 2048 \| Long-range topological similarity \| Slower than ECFP; harder to interpret \|
26	\| TopologicalTorsion \| 4-atom torsion \| All TT \| 2048 \| Path-pattern similarity \| Like AP, slower than ECFP \|
27	\| Avalon \| Substructure + atom pairs \| Mixed \| 512/1024 \| Fast similarity \| Less standard; older \|
28	\| MAP4 (MinHashed atom-pair) \| MinHash atom-pair \| r=1,2 \| 1024/2048 \| Biological + metabolite diversity \| `map4` library required; slower hash \|
29	\| MHFP6 (MinHash) \| MinHash ECFP-like \| r=3 (diam 6) \| 2048 \| Large-library nearest-neighbor with a compatible MinHash/LSH index \| Different distance semantics from folded-bit Tanimoto \|
30	\| Pharm2D \| 2D pharmacophore \| feature pairs/triplets \| sparse \| Pharmacophore search \| Sparse, slower \|
31
32	Decision: For drug-like similarity ranking, start with ECFP4 2048 bit because it is fast and well characterized. MHFP6 outperformed ECFP4 for analog recovery in the benchmark reported by Probst and Reymond (2018), making it a candidate for large, diverse libraries. For scaffold hopping, benchmark ECFP4, AtomPair, TopologicalTorsion, and pharmacophore fingerprints on target-relevant actives and decoys; published comparisons do not support a universal AtomPair advantage (Gardiner et al. 2011; Riniker & Landrum 2013).
33
34	## Bit vs Count Vectors
35
36	\| Form \| Use \| Library impact \|
37	\|------\|-----\|----------------\|
38	\| Bit (0/1) \| Tanimoto similarity, BulkTanimotoSimilarity, RDKit fingerprint folding \| Standard for similarity \|
39	\| Count (integer) \| Some ML methods, RF on counts, neural fingerprints \| Loses bit-level fast operations; richer signal \|
40	\| Sparse (dict) \| Direct chemical interpretation (which fragments at which atoms) \| Use for SHAP / atomic attribution \|
41
42	```python
43	from rdkit import Chem
44	from rdkit.Chem import rdFingerprintGenerator
45
46	mol = Chem.MolFromSmiles('CCO')
47
48	morgan = rdFingerprintGenerator.GetMorganGenerator(radius=2, fpSize=2048)
49	ecfp4_bit = morgan.GetFingerprint(mol)
50	ecfp4_count = morgan.GetCountFingerprint(mol)
51	ecfp4_sparse = morgan.GetSparseCountFingerprint(mol)
52	```
53
54	## Morgan / ECFP Radius Math
55
56	ECFP-X notation: X is the diameter in bonds. RDKit's `radius` parameter is half of X.
57
58	\| Notation \| RDKit radius \| Diameter \| Captures \|
59	\|----------\|--------------\|----------\|----------\|
60	\| ECFP0 \| 0 \| 0 \| Atom identity only \|
61	\| ECFP2 \| 1 \| 2 \| Atom + immediate neighbors \|
62	\| ECFP4 \| 2 \| 4 \| Atom + 2-bond environment \|
63	\| ECFP6 \| 3 \| 6 \| Atom + 3-bond environment \|
64
65	Trade-off: Larger radius captures more specific local environments but increases collisions at fixed `

People who liked this also viewed

6 picks

Type

molecular-descriptors

1.1kstars185forks

View on GitHub

How to install molecular-descriptors?

gptomics/bioskills/molecular-descriptors

$npx -y skills add GPTomics/bioSkills --skill molecular-descriptors

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	## Version Compatibility
2
3	Reference examples tested with: RDKit 2024.09+, numpy 1.26+, pandas 2.2+, map4 1.1+ (MAP4), mhfp 1.9+. Use `mapchiral` separately when the stereochemistry-aware MAP4C fingerprint is intended.
4
5	Before using code patterns, verify installed versions match. If versions differ:
6	- Python: `pip show <package>` then `help(module.function)` to check signatures
7
8	If code throws ImportError, AttributeError, or TypeError, introspect the installed
9	package and adapt the example to match the actual API rather than retrying.
10
11	# Molecular Descriptors
12
13	Featurize molecules for similarity search, QSAR, virtual screening, or ML. Fingerprint performance is dataset- and objective-dependent: ECFP4 is a strong drug-like baseline, atom-pair and topological-torsion fingerprints expose longer-range topology, MAP4/MHFP6 target broader chemical-space searches, and 3D conformer-based descriptors are needed when shape and stereochemistry matter.
14
15	For canonicalization before featurization, see `chemoinformatics/molecular-standardization`. For 3D-only descriptors, see `chemoinformatics/conformer-generation`.
16
17	## Fingerprint Taxonomy
18
19	\| Fingerprint \| Type \| Radius/Path \| Bits \| Use case \| Fails when \|
20	\|-------------\|------\|-------------\|------\|----------\|------------\|
21	\| Morgan (ECFP) \| Circular \| r=2 (ECFP4), r=3 (ECFP6) \| 2048 typical \| Drug-like similarity, ML default \| Loses long-range topology; bit collisions at low nBits \|
22	\| FCFP \| Functional Morgan \| r=2 default \| 2048 \| Pharmacophore-aware similarity \| Same caveats as ECFP; less specific \|
23	\| MACCS \| Substructure key \| 166 fixed bits \| 167 \| Quick fingerprint, drug-likeness \| Too sparse for large diverse libraries \|
24	\| RDKit FP \| Path/subgraph-based \| paths and branched subgraphs up to 7 bonds by default \| 2048 \| RDKit-native ECFP alternative \| Drug-like only; not optimal for scaffold hopping \|
25	\| AtomPair \| Pair + topological distance \| All atom pairs \| 2048 \| Long-range topological similarity \| Slower than ECFP; harder to interpret \|
26	\| TopologicalTorsion \| 4-atom torsion \| All TT \| 2048 \| Path-pattern similarity \| Like AP, slower than ECFP \|
27	\| Avalon \| Substructure + atom pairs \| Mixed \| 512/1024 \| Fast similarity \| Less standard; older \|
28	\| MAP4 (MinHashed atom-pair) \| MinHash atom-pair \| r=1,2 \| 1024/2048 \| Biological + metabolite diversity \| `map4` library required; slower hash \|
29	\| MHFP6 (MinHash) \| MinHash ECFP-like \| r=3 (diam 6) \| 2048 \| Large-library nearest-neighbor with a compatible MinHash/LSH index \| Different distance semantics from folded-bit Tanimoto \|
30	\| Pharm2D \| 2D pharmacophore \| feature pairs/triplets \| sparse \| Pharmacophore search \| Sparse, slower \|
31
32	Decision: For drug-like similarity ranking, start with ECFP4 2048 bit because it is fast and well characterized. MHFP6 outperformed ECFP4 for analog recovery in the benchmark reported by Probst and Reymond (2018), making it a candidate for large, diverse libraries. For scaffold hopping, benchmark ECFP4, AtomPair, TopologicalTorsion, and pharmacophore fingerprints on target-relevant actives and decoys; published comparisons do not support a universal AtomPair advantage (Gardiner et al. 2011; Riniker & Landrum 2013).
33
34	## Bit vs Count Vectors
35
36	\| Form \| Use \| Library impact \|
37	\|------\|-----\|----------------\|
38	\| Bit (0/1) \| Tanimoto similarity, BulkTanimotoSimilarity, RDKit fingerprint folding \| Standard for similarity \|
39	\| Count (integer) \| Some ML methods, RF on counts, neural fingerprints \| Loses bit-level fast operations; richer signal \|
40	\| Sparse (dict) \| Direct chemical interpretation (which fragments at which atoms) \| Use for SHAP / atomic attribution \|
41
42	```python
43	from rdkit import Chem
44	from rdkit.Chem import rdFingerprintGenerator
45
46	mol = Chem.MolFromSmiles('CCO')
47
48	morgan = rdFingerprintGenerator.GetMorganGenerator(radius=2, fpSize=2048)
49	ecfp4_bit = morgan.GetFingerprint(mol)
50	ecfp4_count = morgan.GetCountFingerprint(mol)
51	ecfp4_sparse = morgan.GetSparseCountFingerprint(mol)
52	```
53
54	## Morgan / ECFP Radius Math
55
56	ECFP-X notation: X is the diameter in bonds. RDKit's `radius` parameter is half of X.
57
58	\| Notation \| RDKit radius \| Diameter \| Captures \|
59	\|----------\|--------------\|----------\|----------\|
60	\| ECFP0 \| 0 \| 0 \| Atom identity only \|
61	\| ECFP2 \| 1 \| 2 \| Atom + immediate neighbors \|
62	\| ECFP4 \| 2 \| 4 \| Atom + 2-bond environment \|
63	\| ECFP6 \| 3 \| 6 \| Atom + 3-bond environment \|
64
65	Trade-off: Larger radius captures more specific local environments but increases collisions at fixed `

People who liked this also viewed

6 picks

Type