About This Project

Multi-omic data (e.g., genomics, transcriptomics) holds the key to identifying novel solutions for most pressing problems in biomedical sciences, such as cancer precision therapy [1]. Large amount of this data exists in databases, but remains unanalyzed [2].

We hypothesize that extraction and integration of these multi-omic features across major cancer types will reveal novel molecular signatures that can improve treatment response or disease progression.

Ask the Scientists

What is the context of this research?

Advances in next-generation sequencing (NGS) methods (e.g., genomics, transcriptomics) enabled a range of cancer omic analyses at the molecular level [1]. These methods identify features underlying cancer pathophysiology and serve pivotal to devise therapeutics [2]. Integrated analysis of multiple omic features can suggest optimal treatment flow at a patient level - termed precision medicine [3].

To get insights, patient omic data is matched with existing omic patterns accumulated through years of research [4]. A major hurdle here is that a large amount of omic data with targetable, but yet unanalyzed features exists in several databases. Also, this data is heterogenoeous, requires complex analysis steps, not validated for clinical actionability, and lacks a multi-omic view [5].

We hypothesize that integrating 15 key omic features across major cancer types will reveal molecular patterns that can guide more effective, data-driven approaches to cancer precision therapy.

What is the significance of this project?

Multi-omic view of disease progression holds the key to identifying novel solutions for some of the most pressing problems in healthcare and biomedical sciences, such as cancer precision therapy, and finding new drug targets or biomarkers [1].

Rapid advances in DNA/RNA/protein sequencing methods made it possible to obtain complete molecular profile of a patient [2]. However, the existing multi-omic data that is required to match with patient data remain largely unanalyzed and scattered [3]. This omic data is large, heterogeneous, and importantly extraction of meaningful insights require specific technical skills, scientific know-how, and computational infrastructure, which is largely lacking among primary stakeholders [4].

This project will provide a proof-of-concept for how a user-facing, AI/ML driven tool encompassing pre-analyzed, pre-validated clinically actionable omic features can drive deeper and faster biological insights.

What are the goals of the project?

Completion of this project would provide a proof-of-concept and identify novel multi-omic, oncologic patterns that correlate with clinical outcomes of cancer patients.

The outcome from this work is also critical for our larger goal to develop a disease-wide Multi-omic Tensor for 50+ omic features corresponding to 85+ human disease conditions (>2Bi data points). Our eventual goal is to combine this Multi-omic Tensor with AI/ML driven omic data analysis kits in cloud to build world's first GPS for biomedical sciences.