Deep Learning Architectures for Single-Cell Genomics Data

Course Motivation

The discovery of the cells changed our perspective on the underlying structure of tissues and organs. We started to understand that alterations in the cells of an organism could have a causal link to diseases, and we started to catalog human diseases accordingly.

Now, we have in our hands, new high-throughput technologies enable us to unravel the underlying genetic and epigenetic regulation of the biological systems, coupled with imaging and spatial information at the “single-cell” resolution.

These technologies generated a huge and high-dimensional data which gave us exciting challenges to develop computational approaches for data analyses and integration.

This course will be an exciting learning journey to the data science of the single-cell genomics. You will learn about the state of the art deep learning architectures which have been recently developed in our field. This course will help you to gain the know-how skills and knowledge to be part of this new and revolutionary understanding of the biological systems in health and disease.

Lectures and Practicum Planning

  • 12 Weeks

  • Language: English

Pre-requisites

  • Prior experience and knowledge about NGS data-analysis.

  • Deep knowledge of a programming language (R or Python).

  • Basic understanding of machine-learning and statistical-inference models.

Course Objectives

  1. Overview of the current technological advances in the single-cell genomics field.

  2. Deep theoretical understanding for the state of the art computational methods and deep learning algorithms for analyzing the single cell genomics data.

  3. Hand-on and practical-experience on analyzing and integrating single-cell genomics data.

  4. Developing practical skills and intuition for data-interpretation through the realm of different biological contexts.

Content

Week 1

Lecture: History and Motivation.

Week 2

Lecture: Introduction into the single-cell genomics technologies.

Practicum: Single-cell genomics data-structure and formats.

Week 3

Lecture: [Part-1] Single-cell genomics data analyses algorithms.

Practicum: Analyzing single-cell RNA-seq data (clustering and cell type annotations).

Week 4

Lecture: [Part-2] Single-cell genomics data analyses algorithms.

Practicum: Trajectory inference for the hematopoietic-system.

Week 5

Lecture: Deep learning architectures: an energy-based models perspective.

Practicum: Cell-cell communications and cross-talks inference for the tumor microenvironment.

Week 6

Lecture: Deep generative models for single-cell genomics data.

Practicum: Analyzing single-cell ATAC-seq data.

Week 7

Lecture: Graph Neural Network (GNN): connecting the dots.

Practicum: Analyzing single-cell proteomics data.

Week 8

Lecture: DL architectures for single-cell spatial genomics data.

Practicum: Analyzing single-cell spatial-transcriptomic data.

Week 9

Lecture: Novel DL architectures for integrating large-scale multi-omics single-cell data atlases.

Practicum: Muti-assays and multi-omics data integration (single-cell RNA-seq and ATAC data).

Week 10

Lecture: Deep Reinforcement Learning (DRL): AI agents constructing causal paths.

Practicum: Muti-assays and Multi-omics data integration (single-cell RNA-seq and spatial transcriptomics data).

Week 11 & 12

Working Teams: Presenting research projects.