Multi-Omics and Machine Learning: The New Age of Predicting Cancer Drug Response

Multi-Omics and Machine Learning: The New Age of Predicting Cancer Drug Response

February 1, 2022

Bringing in Multi-Omics

For well-established biomarkers in cancer treatment, it may seem unusual to revisit their ability to predict patient response. However, in the age of multi-omics, revisiting these established biomarkers can yield even better predictions of patient outcomes.

This is what Sammut from the University of Cambridge did with HER2-targeted breast cancer therapy. The authors collected genomic, transcriptomic, and clinical profiles of pre-treatment breast tumor biopsies from 168 patients subsequently treated with HER2-targeted therapy and surgery.

Multi-Omic Markers of Breast Cancer Therapy Response

After correlating the outcome at the time of surgery to the multi-omic profiles of the tumours, the researchers found that the pre-treatment tumor environment dictated the treatment response.

The features of the tumor environment that corresponded with incomplete response to treatment were things like the mutational burden of the tumours, gene copy number variations, immune cell infiltration and T cell dysfunction.

Machine Learning for Multi-Omics

One of the challenges of cancer diagnosis and prognosis is the complex interactions between the different features like those identified in the tumor biopsies. This is where multi-omics is particularly useful because it provides a way to measure multiple aspects of tumor biology simultaneously.

This then creates another challenge: how can all of these complex data be integrated into a single predictive model? Machine learning has emerged as a solution for developing robust multi-omic predictive models.

Predictive Power

In this study, Sammut created a multi-omic machine learning model to predict response to HER2 breast cancer treatment, which successfully predicted the level of response in an external validation cohort of 75 patients.

Using multi-omics and machine learning in combination captures and integrates a more total picture of the tumor environment and how that predicts therapy response. Using machine learning while developing multi-omic-based predictive models can – and likely – will be used to develop models for other cancers in addition to breast cancer.

Multi-omics studies generate exponential amounts of data, which requires high-quality, scalable storage plus expert processing and analysis. Don’t have these resources in-house? Reach out to us at Bridge Informatics for a free discovery call to learn how we can help you with your data storage infrastructure, analysis and pipeline development.

Jane Cook, Journalist & Content Writer, Bridge Informatics

Jane is a Content Writer at Bridge Informatics, a professional services firm that helps biotech customers implement advanced techniques in management and analysis of genomic data. Bridge Informatics focuses on data mining, machine learning, and various bioinformatic techniques to discover biomarkers and companion diagnostics. If you’re interested in reaching out, please email or


Recent Posts