On 03/11/2022 00:52, Brianna Greenstein wrote:
Hi, I had some questions about Morgan fingerprint counts. I used
AllChem.GetHashedMorganFingerprint(mol, 2, nBits=2048) to get counts
as descriptors for ML models. I am looking at the feature importance
and some of these bits came up as important. I had a few questions on
understanding these hashed fingerprints.
I suspect the Hashed version is a bitstring.
The fact that you also pass nBits also hints towards a bitstring.
If you want the counted version of this fingerprint, use something like
AllChem.GetMorganFingerprint(mol,2).
I am not competent to answer your other questions below.
Last time I wanted to investigate which fp bit means what in a molecule,
I ended
up reimplementing a counted circular fingerprint myself.
I do like rdkit, but its API is quite convoluted sometimes.
* Are the structures the bits represent the same for
GetHashedMorganFingerprint and GetMorganFingerprintasBitVect?
* How can I visualize what a specific bit in the hashed morgan
fingerprint looks like? Can I use DrawMorganBit to visualize it the
same way I would for the normal fingerprint?
_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
