Please help for a work

[eamartinez]

Hi:
I am researching in computational chemistry, on creating an automated workflow in KNIME, for the generation of new QSAR models based on deep neural networks that predict antimalarial activity. But I would like to carry out a more in depth study on my database, for this I would like to carry out a SAR study (structure-activity relationship), to see, the main chemotypes, the presence of activity cliffs (very similar structures but with a great difference in their potency, they represent discontinuity points in a SAR study), among other things. This study is carried out very easily in the SARANEA software, but it would be much more convenient to implement it in KNIME. I have been going through the literature and there is not much about this in KNIME, I found some nodes (activity cliffs viewer, Fingerprint Similarity, Similarity viewer) that could be useful for this purpose but they don’t work as expected. Could someone help me with this?
Thank you very much in advance,

[elsamuel]

Hi @eamartinez,

Have you looked at this SAR example on the KNIME Hub?

Workflow was designed as part of the 5th KNIME UGM Meeting 1-3rd February...

[aworker]

Hi Elier,

The example mentioned by @elsamuel is interesting but please be aware that it is obsolete and you will need to update it to latest version of KNIME. For instance, New indigo nodes do not require to transform data into specific format anymore. Therefore, nodes such as “Molecule to Indigo” are not anymore needed, among other changes you would need to do too.

Besides this, you are looking for examples similar to what SARANEA does (https://pubs.acs.org/doi/pdf/10.1021/ci900416a). I would hence recommend to have a look at the following KNIME workflows by Greg Landrum:

• Workflow illustrating Relational Networks (similar to SARANEA) -> https://kni.me/w/EnPGvhC-KPxErvRq and its pdf presentation (http://cheminformatics.epfl.ch/workshop/2018/landrum.pdf)
• A presentation by GL about KNIME & MALARIA : https://www.youtube.com/watch?v=mCDaS7dTPrQ
• Greg Landrum gave a tutorial on workflows developed for ligand-based virtual screening, based on results of a phenotypic HTS against malaria.
  → Workflows and data from Greg’s presentation : https://kni.me/w/Tnvjf8cpUq03Cegf
  → Greg’s slides : https://www.slideshare.net/GregLandrum1/processing-malaria-hts-results-using-knime-a-tutorial

Hope all this is of help to you

Suerte y exito !

Ael