Lecture 1.3: AI/ML Across Biological Scales | Where Computation Meets Biology

Описание: In this lecture, we examine where computation meets biology across multiple biological scales, from individual atoms and molecules to cells, tissues, and entire patient populations. The goal is to understand how different biological scales produce fundamentally different data structures, and how those structures determine which machine learning approaches are most appropriate.

Rather than treating drug discovery as a set of disconnected steps, this lecture frames it as a continuous multi-scale data pipeline in which information flows from molecular design to clinical outcomes.

In This Lecture We Cover

Biological Scales

Molecular and macromolecular levels
Cellular and tissue-scale biology
Clinical and population-level data

Data and Model Alignment

Molecules (graphs, SMILES) → Graph Neural Networks (GNNs)
Proteins and RNA (sequences, structures) → Transformer models
Cells and tissues (images, omics data) → CNNs and Vision Transformers
Clinical and operational data (tables, text) → Tree-based models, time-series models, and NLP

The Connected Pipeline

How molecular, cellular, and clinical data inform one another
Why effective therapeutic modeling requires cross-scale thinking

Series Context

This course is designed to equip AI and machine learning engineers with the biological intuition and domain knowledge needed to build models for modern therapeutics. The focus is not only on algorithms, but also on understanding the biological systems and data structures those algorithms must operate on.