Speaker Profile
Ph.D., CSO, Genomatix
Biography
After finishing his PhD at the Technical University of Munich and a year as visiting scientist at Brunel University in London, Dr. Martin Seifert joined Genomatix as VP of business development and consulting in 2004, adding his expertise in chromatin IP, transcriptome analysis, and nuclear receptors to the company’s bioinformatics efforts. In his current role of CSO, Dr. Seifert is focused on translational bioinformatics, including how multi-scalar analyses that integrate genomics, epigenomics, gene expression, and phenotype data can be applied to Precision Medicine. Dr. Seifert has over 40 peer-reviewed publications, and has received multiple national and international awards.
Talk
AI and Data Sciences Showcase: New Insights – Machine Learning and Computational Biology
Machine learning is used to improve the performance of a specific task, without having to be explicitly programmed. This technology can be used to build classifiers that separate samples into two or more classes or groups depending on their associated data. There are multiple algorithms available for this. In addition to their use as a classifier for new data, this training process can also be used to identify features within the data that work as separators between the groups. Assuming that these separators mirror the underlying biological differences between the classes, the knowledge gained can help one better understand the molecular processes at work.
Here we show strategies to apply machine learning technologies to DNA Methylation data from three cohort studies to elucidate the effect of smoking on DNA methylation, and use computational biology to look into the associated genetic mechanisms. Combining this information with existing biological knowledge, one can not only find useful classifiers, but also garner a deep insight into the biology at work and its potential clinical implications.
Thus, using machine learning in combination with a good understanding of data classes, and a projection of this information onto biomedical knowledge (e.g. pathways) are key elements to uncover and understand causalities of biological mechanisms.
