The chemical structure must fullfil the following requirements:

• Structure should be uncharged or charges should be balanced

• Only single, double and triple bonds are allowed

• Structure should contain at least 3 carbon atoms

• Structure should contain only one component, single atoms are not considered

• Absolute molecular weight should exceed 1250

General information:

Proteins of the human body constitute large and important subset of the therapeutic targets, here is the tool to predict human protein targets, which could be affected by the given chemical structure, freely available version of the PASS Targets [1]:

• Chemical and biological data were extracted from ChEMBL v36 [2].

• Using these data classifiers were build using PASS software [3].

• In addition to the PASS prediction, summary of the protein classes (as defined in ChEMBL) and GO-terms (slim version) [4] describing biological processes are provided to the users.

Ongoing update from february 2026: RDKit.JS is now used for the intial chemical structure processing [5] instead of the MarvinJS (https://chemaxon.com/). Also, the code is available via GitHub:

https://github.com/pavelVPo/IBMC_LSFBD_webApp_PASS_human_protein_targets__2026

This app allows user to predict interaction with the human protein targets. The scores for each compound is expressed as a difference between probabilities for chemical compound to interact and to not to interact with the particular protein target, which are computed using PASS software based on the data from ChEMBL.

Only protein targets with Pa > Pi (score > 0) are considered as possible for a particular compound and provided to the user.

The higher the scores, the higher the chances for the compound to be found active in the experiment against this particular protein.

I.e., interpretation of the results is quite simple: the higher the scores - the higher the chances for the compound to be found active in the experiment. Detailed explanation of how to interpet the results of PASS is given in this publication

Nevertheless, it should be noted that 3 distinct scores are provided to the users in the current version of PASS human Protein targets:

- Propensity score allows to assess whether chemical compound described by the given structure has propensity to be tested against the particular protein target (please, see this publication). Data on 438888 compounds was used to build classifers having mean IAP (numerically equal to ROC AUC) about 0.97;

- D score allows to assess potential activity of the compound described by the given structure based on the ChEMBL data obtained in simple biosystems including target, chemical compound under study and some reporter (please, see this code and BioAssay Ontology (BAO)) Data on 127491 compounds was used to build classifers having mean IAP (numerically equal to ROC AUC) about 0.96;

- I score allows to assess potential activity of the compound described by the given structure based on the ChEMBL data obtained in more complex biosystems, where indirect influence of the compound on the target's activity is possible via direct interactions with partners of the main target (please, see this code and BioAssay Ontology (BAO)). Data on 167155 compounds was used to build classifers having mean IAP (numerically equal to ROC AUC) about 0.98;

------------------------------

Only the results of the prediction for the targets having at least 10 studied compounds are provided to the user. Full list of the activities, for which D-scores could be computed is provided here and full list of the activities, for which I-scores could be computed is provided here

Laboratory for Structure-Function Based Drug Design, Department for Bioinformatics, Institute of Biomedical Chemistry (IBMC) Pogodinskaya Str. 10, Moscow, Russia, 119121

------------------------------

• Way2Drug Team Tel: +7 499 246-09-20 Fax: +7 499 245-08-57 E-mail: pass@ibmc.msk.ru

• pogodinpv@ibmc.msk.ru

• pogodinpv@gmail.com