Search with MadFast

Yesterday I discovered MadFast. I am very favorably impressed with its scope and speed. However, is it possible to tune a search so that it will pick up analogs that have isoelectronic structures with different elements? For example, when I search for compounds similar to diisopropylphosphorofluoridate (DFP), all the hits had P-O-R structures, but none of them had P-NH-R structures. Thank you.

Dear Rudy,

we are glad that our MadFast sparked an interest in you. However, IJC team is not really qualified yet to answer you competently, so I forwarded your question to the developer responsible. He should contact you either here, via forum, or directly via email.

Wishing all the best
David 

Hi,

I did not know where I should post my question about MadFast, as there appears to be no sub-forum for it as yet.

Thank you for forwarding my question to the developer.

-- Rudy

 

Dear Rudy,

Currently our ECFP/FCFP implementation has the capabilities to provide such tuning. This setting is not exposed in the current version of MadFast however i plan to implement it soon.

I would recommend to see the ECFP documentation at

https://docs.chemaxon.com/display/docs/Extended+Connectivity+Fingerprint+ECFP

And the default configuration file example at https://docs.chemaxon.com/download/attachments/41129785/ecfp.xml?version=1&modificationDate=1409137984000&api=v2 for examples on the available built-in properties.

As a different approach a custom canonicalization (standardization) on the input structures before fingerprint calculation could also be considered: changing every atom to a selected (canonical) isoelectronic one. If this approach seem usable i can check for a possible configuration point for implementing it.

regards,

Gabor

Hi Gabor,

Thanks for your rapid response. I look forward to further developments!

Best regards,

Rudy