The applicability domain of machine learning models trained on structural fingerprints for the prediction of biological endpoints is often limited by the lack of diversity of chemical space of the training data. In this work, we developed “similarity-based merger models” which combined the output of individual models trained on cell morphology (based on Cell Painting) and chemical structure (based on chemical fingerprints) and the structural and morphological similarities of the test compounds to training compounds. We applied these similarity-based merger models using logistic equations to weigh individual features and predicted assay hit calls of 177 assays from ChEMBL, PubChem and the Broad Institute, where the required Cell Painting annotations were available. We found that the similarity-based merger models outperformed other models with an additional 20% assays (79 out of 177 assays) with an AUC>0.70 compared with 65 out of 177 assays using structural models and 50 out of 177 assays using Cell Painting models. Our results demonstrate that similarity-based merger models combining structure and cell morphology models can more accurately predict a wide range of biological assay outcomes and expand the applicability domain by better extrapolating to new structural and morphology spaces.

Datasets: Cell Painting Descriptors (before feature selection) for all compounds available from the BROAD Cell Painting assay: Cell_Painting_Median_Features_Median_doses.csv md5:635daafefd295b1a48aeb19fc7674d55 1.1 GB Selected 89 assays, Cell Painting Descriptors and Morgan Fingerprint bits (after feature selection) for 15272 compounds in the BROAD dataset: BROAD_assay_data_filtered_89_inchi.csv md5:527e7ee2b58ad7a0b0194541920c0f92 5.5 MB BROAD_CP_filtered_15272.csv md5:4325cfe9cf66e38b2bba960eac1e3336 59.3 MB BROAD_Mfp_filtered_15272.csv md5:65905e26c301e3d7e598e89dc2f0759f 11.3 MB Selected 88 assays, Cell Painting Descriptors and Morgan Fingerprint bits (after feature selection) for 9876 compounds in the Public dataset: Public_assay_data_filtered_88_inchi.csv md5:a52895a6a8480d25343d7e7b6fdbcc24 2.8 MB Public_CP_filtered_9876.csv md5:2606037db1d9e22044454bd9f1eb41a0 36.3 MB Public_Mfp_filtered_9876.csv md5:846931cd6c9eac29caff424946a63d5e 4.4 MB

{"references": ["Bray, M. A.; Gustafsdottir, S. M.; Rohban, M. H.; Singh, S.; Ljosa, V.; Sokolnicki, K. L.; Bittker, J. A.; Bodycombe, N. E.; Dan\u010d\u00edk, V.; Hasaka, T. P.; Hon, C. S.; Kemp, M. M.; Li, K.; Walpita, D.; Wawer, M. J.; Golub, T. R.; Schreiber, S. L.; Clemons, P. A.; Shamji, A. F.; Carpenter, A. E. A Dataset of Images and Morphological Profiles of 30 000 Small-Molecule Treatments Using the Cell Painting Assay. Giga-Science. 2017, pp 1\u20135.", "Moshkov, N.; Moshkov, T.; Yang, K.; Horvath, P.; Dancik, V.; Wagner, B. K.; Clemons, P. A.; Singh, S.; Carpenter, A. E.; Caicedo, J. C. Predicting Compound Activity from Phenotypic Profiles and Chem-ical Structures. bioRxiv 2022, 2020.12.15.422887"]}