bblean.fingerprints

Utilites for manipulating fingerprints and fingerprint files

Functions

make_fake_fingerprints	Make random fingerprints with statistics similar to (some) real databases
fps_from_smiles	Convert a sequence of smiles into chemical fingerprints
pack_fingerprints	Pack binary (only 0s and 1s) uint8 fingerprint arrays
unpack_fingerprints	Unpack packed uint8 arrays into binary uint8 arrays (with only 0s and 1s)

bblean.fingerprints.make_fake_fingerprints(num, n_features=2048, pack=True, seed=None, dtype=<class 'numpy.uint8'>)

Make random fingerprints with statistics similar to (some) real databases

bblean.fingerprints.fps_from_smiles(smiles, kind='ecfp4', n_features=2048, dtype=<class 'numpy.uint8'>, sanitize='all', skip_invalid=False, pack=True)

Convert a sequence of smiles into chemical fingerprints

bblean.fingerprints.pack_fingerprints(a)

Pack binary (only 0s and 1s) uint8 fingerprint arrays

bblean.fingerprints.unpack_fingerprints(a, n_features=None)

Unpack packed uint8 arrays into binary uint8 arrays (with only 0s and 1s)

Note

If n_features is not passed, unpacking will only recover the correct number of features if it is a multiple of 8, otherwise fingerprints will be padded with zeros to the closest multiple of 8. This is generally not an issue since most common fingerprints feature sizes (2048, 1024, etc) are multiples of 8, but if you are using a non-standard number of features you should pass n_features explicitly.