Deep Learning in Genomics and Biomedicine
Course Overview
Recent breakthroughs in high-throughput genomic and biomedical data are transforming biological sciences into "big data" disciplines. In parallel, progress in deep neural networks are revolutionizing fields such as image recognition, natural language processing and, more broadly, AI. This course explores the exciting intersection between these two advances. The course will start with introduction to deep learning and overview the relevant background in genomics and high-throughput biotechnology, focusing on the available data and their relevance. It will then cover the ongoing developments in deep learning (supervised, unsupervised and generative models) with the focus on the applications of these methods to biomedical data, which are beginning to produced dramatic results. In addition to predictive modeling, the course emphasizes how to visualize and extract interpretable, biological insights from such models. Recent papers from the literature will be presented and discussed. Students will work in groups on a final class project using real world datasets.
Prerequisites
College calculus, linear algebra, basic probability and statistics such as CS109, and basic machine learning such as CS229. No prior knowledge of genomics is necessary.
Lecture Venue and Times
09/25/2017 - 12/09/2017 Mon, Wed 3:00 PM - 4:20 PM at Hewlett Teaching Center 201
Recitation. Fridays 10:30am - 11:20am at 380-380D
Instructors
Anshul Kundaje (Links to an external site.), Assistant Professor (akundaje@stanford.edu)
James Zou (Links to an external site.), Assistant Professor (jamesz@stanford.edu)
Office hours
James Zou (Links to an external site.): Mondays 4:30-6pm (Packard 253).
Anshul Kundaje: Mondays 4.30-5.30 pm (Lane Med School Building L301)
Assignments
Course project (60%): the students will form teams of 3-5 and choose from one of the suggested projects or select their own project. Teams will be given Microsoft Azure credits to implement algorithms and perform analysis. Teams are expected to work on the research project throughout the second half of the quarter and produce conference-style papers. Each team will present the paper to the entire class at the end of the semester. The course project will consists of the following milestones:
- Project proposal in class (3 minute talk).
- First draft of the paper for peer review.
- Poster presentation (12/11 3:30-6:30pm).
- Final paper (due at noon on Friday 12/15).
A significant portion of the class will be based on reading and discussing the latest literature. Every student should read the assigned papers before class and participate in discussions.
Paper presentation (20%): each team selects one of the suggested papers to present in detail to the class. The presentation should be 20 mins + 5 mins for Q&A. Each team will also write a concise review of the paper. The review will be published on bioRxiv and the Zou group blog.
Peer project review (10%): each team will be assigned two other groups' paper drafts to review. The review should concisely summarize the key findings of the paper, highlight interesting ideas, weaknesses and give suggestions.
Class participation (10%): every student should actively engage in paper discussions in class.
Schedule
The first few lectures will cover the basics of deep learning---convolutional and recurrent architectures, generative models, and optimization/regularization. We will also study the applications of deep learning in several biomedical domains---genomics, protein structure, imaging and medical records.
Date | Topic | Papers | Recitation topic | Assignment |
9/25 | Overview. Intro to machine learning |
1. Deep Learning (See other readings in Files Section) |
||
9/27 | Genomics |
1. Neural Nets and Deep learning primer (See other readings in Files Section) |
Deep learning primer | |
10/2 | DenseNets + Convolutional Nets for Genomics |
1. Basset: learning the regulatory code of the accessible genome with deep convolutional neural networks http://genome.cshlp.org/content/26/7/990.full 2. Denoising genome-wide histone ChIP-seq with convolutional neural networks |
||
10/4 | Recurrent NN | 1. DanQ: a hybrid convolutional and recurrent deep neural network for quantifying the function of DNA sequences https://www.ncbi.nlm.nih.gov/pubmed/27084946 |
Genomics primer | Projects released on 10/6. |
10/9 | Autoencoders + representation learning | |||
10/11 | Optimization + regularization + Azure Demo |
TensorFlow primer | Select projects and papers. | |
10/16 | Generative models | GANs for Biological Image Synthesis https://arxiv.org/abs/1708.04692 |
||
10/18 | Instructor led paper presentations |
1. Basenji: Sequential regulatory activity prediction across chromosomes with convolutional neural networks |
TensorFlow tutorial 2 | |
10/23 | Interpretation of black-box models |
1. The Mythos of Model Interpretability |
||
10/25 | Paper presentations |
1. Attend and Predict: Understanding Gene Regulation by Selective Attention on Chromatin |
Deep learning review | |
10/30 | Proposal presentations |
|
3 minute proposal presentations | |
11/1 | Guest lecture: Google Brain team |
1. DeepVariant: Creating a universal SNP and small indel variant caller with deep neural networks |
||
11/6 | Drug discovery + protein structure |
1. Deep learning for computational chemistry http://onlinelibrary.wiley.com/doi/10.1002/jcc.24764/abstract |
||
11/8 | Paper presentations |
1. Accurate De Novo Prediction of Protein Contact Map by Ultra-Deep Learning Model |
||
11/13 | Paper presentations |
1. druGAN: An Advanced Generative Adversarial Autoencoder Model for de Novo Generation of New Molecules with Desired Molecular Properties in Silico http://pubs.acs.org/doi/abs/10.1021/acs.molpharmaceut.7b00346 3. Convolutional Networks on Graphs for Learning Molecular Fingerprints |
||
11/15 | Imaging + Electronic Medical Records |
1. Deep learning for healthcare: review, opportunities and challenges https://academic.oup.com/bib/article-abstract/doi/10.1093/bib/bbx044/3800524/Deep-learning-for-healthcare-review-opportunities |
||
11/27 | Paper presentations |
1. Privacy-preserving generative deep neural networks support clinical data sharing |
Initial paper submitted for peer review | |
11/29 | Paper presentations |
1. Learning to Detect Sepsis with a Multitask Gaussian Process RNN Classifier |
Peer review out | |
12/4 | Paper presentations |
1. Learning Sleep Stages from Radio Signals:A Conditional Adversarial Architecture |
||
12/6 | Wrap up | Opportunities/Obstacles Deep Learning Biomedicine https://www.biorxiv.org/content/early/2017/05/28/142760 |
||
12/11 | Finals week: poster presentation | Final paper due 12/15. |
Course Summary:
Date | Details | Due |
---|---|---|