thx for the fast answer.
Viele Grüße
Till
Am Mittwoch, 4. November 2015, 14:09:54 schrieb Egon Willighagen:
> Dear Till,
>
> neither. The default (oldest) CDK fingerprint is like the Daylight
> fingerprint (using the same approach), but is not meant to be
> identical.
>
> Grtz,
>
> Egon
>
Oh, sorry the CDK 1.4 identical to 1.5... No, they are typically
not, as they depend on an aromaticity model. This is why I introduced
the SP2-based fingerprinter, which also takes into account
hybridization, but does not depend on definitions of aromaticity...
and should be more stable.
(But
Hi Till,
Have you found one?
Nothing immediately jumps out from recent history -
https://github.com/cdk/cdk/blob/master/base/standard/src/main/java/org/openscience/cdk/fingerprint/Fingerprinter.java
Further history I see count fingerprints introduced but that shouldn't
change the bits -
Hi,
i would like to know if the daylight fingerprint should be identical for the
latest CDK 1.4 and the latest CDK 1.5. Are there any changes, that cause
another ordering of the features, or some feature to be detected differently?
Regards,
Till
--
Dipl.-Inf. Till Schäfer
TU Dortmund
Yeah, see 2nd reply, as I realized the same thing after John's email...
So, no they are not. In the past we frequently had to reculate FPs for
unit tests, and at some point we made a full test, just to see if the
FP calculation changed :)
So, new CDK version, then you must recalculate
5 matches
Mail list logo
