Grading: Your grade will be based on two parts: course project (75%) and discussion in five guest lectures (5% each, 25% in total). There is no credit given for attending other lectures that are offered by the instructor. There is no exam.
Guest lectures: 45-minute invited presentation about ongoing AI for drug discovery research in other places (see schedule below). The instructor will then lead the discussion about the potential improvement and future directions.
Projects: We focus more on proposing rather than finishing a project. Students are required to do a short mid-term presentation, write formal reviews for others' projects and submit a final project report.
Grading the project: Mid-term presentation (15 points). Reviewing others' projects (30 points). Final project report (30 points). Mid-term presentation and final project will be graded by other students in the class. Instructor and TA will grade the peer-review and calibrate the final score.
4/7. Form a group (group size: TBA)
4/14. Meet with the instructor before this date to decide the project topic. The instructor will suggest readings.
4/28. Mid-term presentation (10 minutes presentation + 5 minutes QA). This presentation needs to cover project idea, literature review and differentiation from previous works. The presentation will be graded by other students based on the novelty.
5/3. Peer-review of mid-term presentation.
5/31 & 6/2. Final project presentation. This presentation needs to cover preliminary results, future plan, potential pitfalls and impact. It will be graded by other students. We don't expect you to finish this project. We only expect some preliminary results.
6/6. Peer-review of the final project report. Criteria: Significance (is this problem important?), Approach (is the proposed method appropriate), Innovation (Is this idea original).
| Date | Theme | Reading/Content in Drug discovery | Reading/Content in Machine learning | Slides |
|---|---|---|---|---|
| AI for Drug Discovery in the industry | ||||
| 3/29 | Overview of drug discovery pipeline. Review of AI for Drug Discovery in Startups | BenevolentAI, Insitro, HeliXonAI, DeepCure, Standigm, CytoReason, GV20 Oncotherapy, Exscientia | Deep learning for DDR1 inhibitors
Another AI Drug Announcement |
|
| 3/31 | Review of AI for Drug Discovery in pharmaceutical companies | Johnson, Roche, Novartis, Merck, Bristol-Myers Squibb, Pfizer, Sanofi, AstraZeneca | Deep learning for cellular image analysis Applications of machine learning in drug discovery and development |
|
| 4/5 | Introducing course project ideas | Exploration of databases and methods supporting drug repurposing: a comprehensive survey Massively multiplex chemical transcriptomics at single-cell resolution. Inconsistency in large pharmacogenomic studies |
||
| Graph-based drug discovery | ||||
| 4/7 | Guest lecture by Dr. Wengong Jin at Broad Institute. | Antibody sequence-structure co-design, ICLR 2022 Drug combinations for treating COVID-19, PNAS 2021 Deep Learning for Antibiotic Discovery, Cell 2020 |
||
| 4/12 | Graph-based drug discovery (part 1) | Modeling polypharmacy side effects with graph convolutional networks Drug repositioning by integrating target information through a heterogeneous network model |
Neural Message Passing for Quantum Chemistry | |
| 4/14 | Graph-based drug discovery (part 2) | ZINC | Strategies for pre-training graph neural networks Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generation |
|
| Structure-based drug discovery | ||||
| 4/19 | In-person Guest lecture by Dr. Minkyung Baek at UW. | RoseTTA fold, Science 2021 |
||
| 4/21 | Structure-based drug discovery (part 1) | Protein Docking, Drug Binding Structure Prediction, Molecular 3D Conformer | E(n) Equivariant Graph Neural Networks EQUIBIND: Geometric Deep Learning for Drug Binding Structure Prediction INDEPENDENT SE(3)-EQUIVARIANT MODELS FOR END-TO-END RIGID PROTEIN DOCKING |
|
| 4/26 | Structure-based drug discovery (part 2) | |||
| Genomics-based drug discovery | ||||
| 4/28 | Student project updates/presentation | |||
| 5/3 | Guest lecture by Dr. Peng Jiang at NIH. | Immunotherapy biomarkers/targets discovery, Nature Methods 2021, Patterns 2022 | ||
| 5/5 | Precision medicine (Part 1) | Multi-omics data integration | (Unpaired) data integration, Batch correction | |
| 5/10 | Precision medicine (Part 2) | Drug repurposing, drug combination therapy, drug side effects, personalized dosage | ||
| 5/12 | Guest lecture by Ruishan Liu at Stanford. | Clinical trials and real-world data, Nature 2021 | ||
| Sequence-based drug discovery | ||||
| 5/17 | NLP for drug discovery (part 1) | Graphine, Pathway2text | ||
| 5/19 | Guest lecture by Xuan Wang, incoming assistant professor at Virginia Tech. | Automated Scientific Knowledge Extraction from Massive Text Data | ||
| 5/24 | Guest lecture by Kevin Yang at Microsoft Research. | Protein sequence engineering. | ||
| 5/26 | Guest lecture by Martin Jinye Zhang at Harvard. | Polygenic enrichment distinguishes disease associations of individual cells in single-cell RNA-seq data. | ||
| 5/31 | Student project presentation (part 1) | |||
| 6/2 | Student project presentation (part 2) |