Machine Learning for Bioengineering and Applied Life Science

Contact person: Dr. Alexander Lukyanov, Scientific Director Artificial Intelligence and Machine learning, Uncertainty Analysis and Quantification, BISC Global

Machine Learning for Bioengineering and Applied Life Science will be offered to bioengineers, bioinformaticians, computational biologists and researchers to introduce practical data analytics, dimension reduction, and machine learning techniques, for a variety of life science and bioengineering applications in transcriptomics, proteomics, and systems biology domains.

This course is designed for an audience with a background in computational biology, bioinformatics, data science and/or applied life sciences. The course will provide an overview of four major categories of machine learning techniques (reduced-order methods, geometric learning, manifold learning, and reinforcement learning) and a data-driven model-free framework.

Case studies will be used to demonstrate how these learning techniques have enhanced research and technology advancements. These application problems will include a data-driven model-free paradigm for complex systems biology problems, reduced-order modeling of single cell data, geometric learning for protein engineering and multi-omics analysis, and reinforcement learning-enabled multiscale biological modeling.

Target Groups: Bioengineers, bioinformaticians, computational biologists, and researchers in applied life sciences with an understanding of fundamental biological principals. Participants must bring their own laptops for the three-day training course.

Training course materials: A course website containing course materials and sample code repositories will be set up prior to the start of the course.

Topics covered:

• Dimension reduction by manifold learning and autoencoders.
• Multi-objective reinforcement learning .
• Reinforcement learning for biophysical properties prediction via regression / pattern recognition.
• Causal graph discovery .
• Interaction of proteins sites with a variety of low-molecular-weight compounds.
• Predicting protein-protein interactions.
• Application of graph neural networks, convolutional graph neural networks to analyze digital pathology image data as well as covariates (genomics, single cell data, age, sex, smoking status, ECOG, etc).
• Uncertainty analysis & quantification.
  (a) single cell discoveries / classification / annotation pipelines .
  (b) Feature importance analysis (i.e., impact of different biomarkers) .
• Reduced-order classification for single cell analysis problems.
• Manifold (single cell driven manifold) learning enhanced data-driven modelling of single cell differentiation.
• Geometric learning on a manifold for predicting functional role of biomarkers.

Takeaway:  In addition to small projects designed to reinforce specific technical concepts in machine learning, you will build a realistic, complete, machine learning understanding of industry applications using case studies commonly occurring within the industry.

Registration: Secure your spot now. Seats are limited, and we accept applicants on a first-come, first-served basis.