Welcome to the web page for Mathematical and Computational Biology in Drug Discovery, the course series running at the Department of Mathematics and Computer Science, University of Basel in the spring semester 2023.
The course is open to all students who wish to learn about principles and techniques of mathematical and computational biology as well as their applications in drug discovery.
Find administrative details about the lecture in the course directory of University of Basel.
Table of content
- Preparation
- Overview
- Assessment
- Syllabus
- Module Zero: Introduction
- Module I: What are drug targets and where to find them?
- Module II: What can we do if there are no good targets?
- Module III: What kind of drug should we develop?
- Module IV: What efficacy and safety profiles can we expect?
- Module V: For which patients will the drug work and how does it work, really?
- Contact
Preparation
To prepare for the course, please (1) check that you have acquired required prior knowledge, (2) make sure that you can commit enough time to the course, (3) note that lectures will be in person, and (4) fill the pre-course survey.
You should foremost be familiar with the content covered by the course series Introduction to Applied Mathematics and Informatics In Drug Discovery that run in fall semesters.
The MCBDD course on average requires about 4-6 hours’ time each week for reading assignments or programming tasks.
Finally, to make the course useful and valuable for you, I invite you again to fill the pre-course survey. Your opinions help to shape the course.
Overview
Time
Lectures take place on Fridays between 12:15 and 14:00 in Seminarraum 05.002 in Spiegelgasse 5, near Schifflände, 4070 Basel. See Syllabus for the topics we plan to cover.
Topics that we shall discuss
We mainly discuss following topics from biology
- Genetics and population genetics
- Genomics and comparative genomics
- Transcriptomics and proteomics
- Chemical biology
- Functional genomics
- Antisense oligonucleotides and antibodies
- Pharmacokinetics and pharmacodynamics
We mainly discuss applications of following mathematical and computational topics:
- Reproducible research tools, such as git, Snakemake/NextFlow, conda, etc.
- Linear models and generalised linear models
- Inference methods, such as the Expectation-Maximization (EM) algorithm, Hidden Markov Models (HMMs), clustering algorithms, Monte-Carlo methods, and variational methods
- Generative models
- Gene network inference
- Machine learning, deep learning, and Gaussian Process
- Mathematical modelling
Course material and licensing
Course material, including lecture notes, slides, and reading material, is shared on the course’s web site, https://www.MCBDD.ch, under the Creative Commons Attribution-ShareAlike 4.0 Interactional License unless otherwise specified.
Assessment
The final grade is given by participation (50%) and offline activities (50%).
Syllabus
Module Zero: Introduction
Module Zero is an introduction to mathematical and computational biology in drug discovery.
- Slides of lecture #1
- Offline activities: see slide page 20
Module I: What are drug targets and where to find them?
This module consists of two lectures: (1) what makes a good drug target , and (2) how to identify, assess, and validate drug targets?
Prior to attending the courses, you can refresh your knowledge in the central dogma of molecular biology and in the human genome by watching the animation film From DNA to protein - 3D by yourgenome, and the film mRNA processing and the spliceosome by WEHI that combines an artist’s impression and simulation.
- Slides of Module I
- Offline activities of Module I. Please finish this and further offline activities before April 1st to get the corresponding credit.
Module II: What can we do if there are no good targets?
Module II discusses about alternatives to target-based drug discovery, in particular phenotypic drug discovery. It includes two lectures: (1) phenotypic screening with chemogenomic libraries, and (2) molecular phenotypic screening based on gene expression.
- Slides of Module II
- Offline activities of Module II: The explanation can be found in the slidesr, see slide number #27. Please finish the activities and submit the results here via Google Form before May the first to get credits.
Module III: What kind of drug should we develop?
Module III considers modality selection from a computational point of view. It includes two lectures: (1) novel small molecules and antisense oligonucleotides, and (2) antibodies, multi-target drugs, and gene- and cell-therapies.
- Slides of Module III
- Offline activities of Module III: question-guided learning of factor analysis. Use your favorite literature programming tools (i.e. Rmarkdown/Jupyter Notebook) to investigate the topic of factor analysis. Use the questions below to guide your learning.
- What is factor analysis?
- What are the relationships between covariance matrix, factor analysis, and principal component analysis (PCA)?
- What do we mean with loadings?
- Why factors are orthogonal to each other? What’s the consequence?
- How can we use factor analysis as a generative model?
- What is the relationship between factor analysis and autoencoder?
- How can you it explain factor analysis to a high-school student?
Please submit your answers, together with your code on GitHub or other software repo, to this Google Form for Module III offline activity.
Module IV: What efficacy and safety profiles can we expect?
Module IV focuses on MoA inference for safety and efficacy profiles of drug candidates. We will mainly computational analysis and impact of single-cell omics data, and explore the potential of proteomics to infer mode of action.
The offline activities of Module IV is to do single-cell RNA-seq analysis yourself with either Python or R. This activity is optional: it does not count to the final grade. I recommend you doing it because you can get first-hand experience analysing high-dimensional, sparse, and noisy biological data. See slides for links to tutorials and courses.
- If you are new to the topic of single-cell RNA sequencing, please follow the PBMC tutorial of Scanpy (python) or the PBMC tutorial of Seurat (R). They both describe the analysis of a peripheral blood mononuclear cell (PBMC) dataset.
- If you have experience with such data already, checkout the NBIS workshop on single-cell sequencing data analysis to cover advanced topics such as spatial transcriptomics and trajectory inference.
Module V: For which patients will the drug work and how does it work, really?
We consider the problem of causal inference in Module V.
There is no offline activity of Module V.
Contact
In case you have further questions, comments, and suggestions about the course, please contact the lecturer, Jitao David Zhang, at jitao-david.zhang@unibas.ch.