# Integrative machine learning approach for identification of new molecular scaffold and prediction of inhibition responses in cancer cells using multi-omics data

## TL;DR

This study uses machine learning and multi-omics data to identify new drug candidates targeting MDM2 in cancer cells and predict their effectiveness.

## Contribution

The novel integration of machine learning with multi-omics and single-cell RNA-seq data to predict drug inhibition responses and identify new molecular scaffolds.

## Key findings

- Idasanutlin showed a robust response in TP53 wild-type cancer cell lines.
- Transcriptional response signatures were identified to understand drug mechanisms and clinical potential.
- Similarity search and molecular docking identified potential lead compounds from the ChEMBL database.

## Abstract

MDM2 (Mouse Double Minute 2), a fundamental governor of the p53 tumor suppressor pathway, has garnered significant attention as a favorable target for cancer therapy. Recent years have witnessed the development and synthesis of potent MDM2 inhibitors. Despite the fact that numerous MDM2 inhibitors and degraders have been assessed in clinical studies for various human cancers, no FDA-approved drug targeting MDM2 is presently available in the market. Researchers have investigated the effects of various drugs, which are involved in cancer therapies with known mechanisms, on well-characterized cancer cell lines. The prediction of drug inhibition responses becomes crucial to enhance the effectiveness and personalization of cancer treatments. Such findings can provide new perceptions aimed at designing new drugs for targeted cancer therapies. In our current insilico work, a robust response was observed for Idasanutlin in cancer cell lines, indicating the drug’s significant impact on gene expression. We also identified transcriptional response signatures, which were informative about the drug’s mechanism of action and potential clinical application. Further, we applied a similarity search approach for the identification of potential lead compounds from the ChEMBL database and validated them by molecular docking and dynamics studies. The study highlights the potential of incorporating machine learning with omics and single-cell RNA-seq data for predicting drug responses in cancer cells. Our findings could provide valuable insights for improving cancer treatment in the future, particularly in developing effective therapies.

Graphical Abstract

The study investigates the effects of Idasanutlin on well-characterized cancer cell lines, targeting eight cancer therapies with known mechanisms. The researchers identified a robust response in TP53 WT cell lines, highlighting the potential of this drug in cancer therapy. This study identified transcriptional response signatures that provide valuable insight into the drug’s potential clinical application and molecular mechanisms. Mutational effects in significant cancers, including prostate, ovarian cancer, and kidney, were noted, emphasizing the importance of understanding molecular pathways. Integrating machine learning with multi-omics and single-cell RNA-seq data offer promising avenues for predicting drug responses in cancer cells and improving cancer therapy.