Using a Multivariate Modeling approach to Encode the Larger differences in the Human Genome | Schiller International University Skip to main content Skip to footer

The human genome is said to be 99% similar between individuals with only a small difference of 1%, which accounts for our differences in behavior, the way we look, hereditary diseases, etc. In his paper titled, DMWAS: Feature Set Optimization by Clustering, Univariate Association, Deep and Machine Learning Omics Wide Association Study for Biomarkers Discovery as  Tested on GTEx Pilot Dataset for Death Due to Heart-Attack, Professor Singh showcases his methodology and findings to uncover the behavioral patterns of hereditary diseases, especially in the case of heart attacks. 

The idea of this research was inspired by some of professor Singh’s previous work which demonstrated a higher correlation of larger variations in DNA to the physical and behavioral attributes of an individual. In this particular piece of work professor, Singh methodically used various statistical and computational strategies to deal with big data in order to reduce computational costs and prioritize important data points. 



Various existing machine learning models were deployed and the Python codebase was made freely available. Professor Singh believes that sharing code is an important step to help with future research and advancement.  

The work was done on GTEx consortium pilot data for the people who died due to heart disease and the relevant DNA biomarkers were presented with high statistical significance. This work is considered groundbreaking as it gives a practical implementation to replace the traditional univariate association such as GWAS (Genome-Wide Association Study) with DMWAS (Deep and Machine learning based omics Wide Association Study) which is a multivariate modeling approach that provides more robust and statistically signification insights. 

As an Associate Professor of IT-based Businesses at Schiller International University, Professor Singh thinks that there is no better way to teach his students than to 'walk the talk', which served as motivation for him to carry out this complex research to demonstrate the power of technology. 

Abhishek N. Singh (Abi) is an 'Associate Professor of Information Systems and Business' at Schiller International University, Heidelberg, where he has been engaged in teaching several courses in the MBA as well as in BBA programs. 

The link to the complete white paper can be found here

Schiller University partner for dual degrees: University of Roehampton London Logo
ACCSC Accreditation Logo
Comunidad de Madrid Accreditation Logo
evalag Accreditation Logo

About the author