2024 Fall CS582 Machine Learning for Bioinformatics

Additional course websites: UIUC Canvas

Course description

In this graduate-level course, we will introduce foundational and state-of-the-art machine-learning approaches to key problems in life sciences and biology, with an emphasis on deep learning for molecular engineering. The course is structured into four main modules: functional genomics, protein language models, geometric DL, and protein structure models. We will cover a wide range of DL architectures (CNN, Transformer, GNN, etc.), generative models (e.g., language models and diffusion) and machine learning techniques (supervised, unsupervised, optimization), and discuss their applications to modeling, understanding, and designing the sequence, structure, and functions of genomes and proteins. Students will learn about both theoretical concepts and applications to biological problems, and gain practical experience through implementing solutions to problem sets, reading papers, and conducting final course project.

Logistics

• Lecture: WF 11:00 - 12:15 in 1302 Siebel Center for Comp Sci
• Office hours: Thursday 3pm-4pm in 3212 Siebel Center for Comp Sci
• TA office hours: F 10:00-10:50 in Siebel Center floor 0 lobby

Grading

Grading will be based upon four problem sets containing either programming questions or written questions (40%), a Course Project (35%), reading assignments (20%), and participation (5%). Attendance in lecture is important as the class moves quickly and you will need to be present. You can use three late days for problem set deadlines (or email the course staff).

Syllabus and schedule

Date	Lecture	Title	Assignment
8/28/2024	Lecture 1	Logistics and intro
8/30/2024	Lecture 2	ML foundations
9/4/2024	Lecture 3	ML foundations
Module I: Functional Genomics
9/6/2024	Lecture 4	DL for functional genomics I
9/11/2024	Lecture 5	CNN/ResNet
9/13/2024	Lecture 6	RNN, LSTM, StateSpaceModel(S4,Mamba,Hyena)	Reading assignment 1
9/18/2024	Lecture 7	Model uncertainty and interpretability	Problem set 1 out
9/20/2024	Lecture 8	DL for functional genomics II
Module II: Protein Language Models
9/25/2024	Lecture 9	Protein sequenece modeling I
9/27/2024	Lecture 10	Transformer and MLM I
10/2/2024	Lecture 11	Transformer and MLM II	Reading assignment 2
10/4/2024	Lecture 12	Protein sequenece modeling II	Problem set 1 due, Finalize project team-up
10/9/2024	Lecture 13	Protein function prediction (sequence based)	Problem set 2 out
10/11/2024	Lecture 14	Protein function optimization (sequence based) I
10/16/2024	Lecture 15	Protein function optimization (sequence based) II
Module III: Intro to Geometric DL
10/18/2024	Lecture 16	GNN
10/23/2024	Lecture 17	Intro to Geometric DL for molecular representation I	Problem set 2 due, Reading assignment 3
10/25/2024	Lecture 18	Intro to Geometric DL for molecular representation II	Problem set 3 out
Module IV: Protein Structure Models
10/30/2024	Lecture 19	Protein structure prediction I
10/1/2024	Lecture 20	Protein structure prediction II	Project proposal due
11/6/2024	Lecture 21	Folding and inverse folding	Reading assignment 4
11/8/2024	Lecture 22	Protein interaction, function prediction
11/13/2024	Lecture 23	Generative models primers: VAE,Diffusion, Flow Matching I	Problem set 3 due
11/15/2024	Lecture 24	Generative models primers: VAE,Diffusion, Flow Matching II	Problem set 4, Reading assignment 5
11/20/2024	Lecture 25	Generative model for protein design I
11/22/2024	Lecture 26	Generative model for protein design II
Final projects
11/27/2024	No lecture	Fall break
11/29/2024	No lecture	Fall break
12/6/2024	Lecture 27	Project Presentations I
12/8/2024	Lecture 28	Project Presentations II

Prerequisites

Linear Algebra, python programming, intro to ML.

Project

This subject has a substantial project component. We recommend (but do not require) working on projects in team of 2-3 students. You are free to choose any problem related to the lectures of the course, and develop a deep learning solution.