Hi Greg,

Yes I understand now, thanks very much for the explanation!


Susan

________________________________
From: Greg Landrum <greg.land...@gmail.com>
Sent: 20 February 2019 10:37:23
To: Susan Leung
Cc: rdkit-discuss@lists.sourceforge.net
Subject: Re: [Rdkit-discuss] Feature map score over 1

Hi Susan,

Feature maps have different possible scoring modes:
https://github.com/rdkit/rdkit/blob/master/rdkit/Chem/FeatMaps/FeatMaps.py#L15

The default mode - All - scores each point in the probe set against every point 
in the feature map, so the maximum possible score is not capped at 1.0
If you switch the scoring mode to Closest or Best, you'll end up with a max 
average score (what is being calculated in that notebook) of 1.0[1]

-greg
[1] The max value of 1.0 is only because in this case all feature weights are 
1.0.


On Wed, Feb 20, 2019 at 10:51 AM Susan Leung 
<susan.le...@st-hildas.ox.ac.uk<mailto:susan.le...@st-hildas.ox.ac.uk>> wrote:

Dear all,


I am using RDKit's FeatMap functionality and I am following the code from the 
blogpost, http://rdkit.blogspot.com/2017/11/using-feature-maps.html

Using Feature Maps<http://rdkit.blogspot.com/2017/11/using-feature-maps.html>
rdkit.blogspot.com<http://rdkit.blogspot.com>
This post provides a short demonstration of how to use the RDKit's feature map 
implementation to score aligned 3D conformations against...


I was wondering why the output from  
fms[0].ScoreFeats(featLists[0])/fms[0].GetNumFeatures() (In [19]) is giving a 
value of 1.0047133039823972 and not 1? Is this a rounding issue?


Thanks,


Susan


_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net<mailto: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

Reply via email to