AI-powered Molecular Modeling

Instructor	Debswapna Bhattacharya (dbhattacharya@vt.edu)
Class meets	Tuesday and Thursday 11:00 am - 12:15 pm at McBryde Hall 204
Office Hours	Debswapna Bhattacharya: Tuesday and Thursday 1:00 pm - 2:00 pm at Torgersen 3120B
Canvas	https://canvas.vt.edu/courses/196265

Description

Welcome to CS 6824: AI-powered Molecular Modeling! This course will survey the emerging field of computational modeling of molecular structures driven by advances in artificial intelligence (AI), with an emphasis on predictive modeling. We will investigate relevant issues from interdisciplinary perspectives of macromolecular modeling and machine learning focusing on:

Learn about machine learning methods applied to problems in molecular modeling
Learn how to critically read and evaluate papers
Learn how to pose research problems and practice oral and written scientific communication skills

Prerequisites
This course is appropriate for graduate students in computer science, computational biology, bioinformatics, and statistics. Familiarity with fundamental concepts in machine learning, statistics, probability and algorithms is expected.

Grading

Homeworks: 30%
Paper presentations: 20%
Midterm project proposal: 15%
Final project: 25%

Final report (15%)
Final presentation (10%)

Class participation: 10%

In-class engagement (5%)
Post-lecture feedback (5%)

Based on the grading breakdown above, each student's final letter grade for the course will be determined based on the ceiling of the final percentage of points earned. The grade ranges are as follows:

A: 93%-100%
A-: 90%-92%
B+: 87%-89%
B: 83%-86%
B-: 80%-82%
C+: 77%-79%
C: 73%-76%
C-: 70%-72%
D+: 67%-69%
D: 63%-66%
D-: 60%-62%
F: Below 60%

Schedule

Note: This schedule is tentative and subject to change.

Week / Day	Date	Presenter	Topic	Readings
1 / T	Aug 27	Debswapna Bhattacharya	Introduction [Slides]	'It will change everything': DeepMind's AI makes gigantic leap in solving protein structures. Nature 2020. AlphaFold2 @ CASP14: "It feels like one's child has left home."
1 / R	Aug 29	Debswapna Bhattacharya	Course Overview [Slides]
2 / T	Sep 3	Debswapna Bhattacharya	Biological Macromolecules [Slides]	L Hunter. Molecular Biology for Computer Scientists
3 / M	Sep 5	Debswapna Bhattacharya	Convolutional Neural Networks: The Inflection Point [Slides]	LeCun Y, Bengio Y, Hinton G. Deep learning. Nature 2015. He et al. Deep Residual Learning for Image Recognition. CVPR 2016.
3 / T	Sep 10	Debswapna Bhattacharya	Convolutional Neural Networks for Single-Chain Protein Structure Prediction: AlphaFold 1 [Slides]	Senior, A.W., Evans, R., Jumper, J. et al. Improved protein structure prediction using potentials from deep learning. Nature 2020. AlphaFold1 CASP13 slides HW1 out
3 / R	Sep 12	Debswapna Bhattacharya	Convolutional Neural Networks for Single-Chain Protein Structure Prediction: trRosetta [Slides]	Yang et al. Improved protein structure prediction using predicted interresidue orientations. PNAS 2020.
4 / T	Sep 17	Debswapna Bhattacharya	Attention and Transformers: The Paradigm Shift [Slides]	Vaswani et al., Attention is All You Need., NeurIPS 2017. HW1 due
4 / R	Sep 19	Debswapna Bhattacharya	Attention and Transformers for Single- and Multi-Chain Protein Structure Prediction: AlphaFold 2 [Slides]	Jumper, J., Evans, R., Pritzel, A. et al. Highly accurate protein structure prediction with Alphafold. Nature 2021. AlphaFold 2 CASP14 slides Tunyasuvunakool, K., Adler, J., Wu, Z. et al. Highly accurate protein structure prediction for the human proteome. Nature 2021. AlphaFold DB Evans et al. Protein complex prediction with AlphaFold-Multimer bioRxiv 2022. HW2 out
5 / T	Sep 24	Debswapna Bhattacharya	Attention and Transformers for Prediction of Protein Structures and Interactions: RoseTTAFold [Slides]	Baek et al. Accurate prediction of protein structures and interactions using a three-track neural network. Science 2021. [paper]
5 / R	Sep 26	Debswapna Bhattacharya	Biological Language Models: The Gold Mine [Slides]	Wu et al. High-resolution de novo structure prediction from primary sequence. bioRxiv 2022. Lin et al. Evolutionary-scale prediction of atomic-level protein structure with a language model. Science 2023. Rives et al. Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences. PNAS 2020. ESM3: Simulating 500 million years of evolution with a language model HW2 due
6 / T	Oct 1	Debswapna Bhattacharya	RNA Structure Prediction in the Post-AlphaFold2 Era [Slides]	Wang et al. trRosettaRNA: automated prediction of RNA 3D structure with transformer network. Nature Communications 2023. Li et al. Integrating end-to-end learning with deep geometrical potentials for ab initio RNA structure prediction. Nature Communications 2023. HW3 out
6 / R	Oct 3	Trevor Norton	Diffusion Models: The Noisy Revolution [Slides]	Ho et al. Denoising Diffusion Probabilistic Models. NeurIPS 2020. Trippe et al. Diffusion probabilistic modeling of protein backbones in 3D for the motif-scaffolding problem. ICLR 2023.
7 / T	Oct 8	Trevor Norton	Diffusion-based Neural Architecture for Biomolecular Interactions: AlphaFold 3 [Slides]	Abramson et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature 2024. AlphaFold Server HW3 due
7 / R	Oct 10	Trevor Norton	Diffusion-based Generalized Biomolecular Modeling and Design: RoseTTAFold All-Atom [Slides]	Krishna et al. Generalized biomolecular modeling and design with RoseTTAFold All-Atom. Science 2024.
Fall Break
8 / T	Oct 15	Luis Lazcano Feedback by Stephen Owesney	Paper Presentation I [Slides]	Xu et al. Improved protein structure prediction by deep learning irrespective of co-evolution information. Nature Machine Intelligence 2021.
8 / R	Oct 17	Luke Elder Feedback by Luis Lazcano	Paper Presentation I [Slides]	Baek et al. Accurate prediction of protein-nucleic acid complexes using RoseTTAFoldNA. Nature Methods 2024.
9 / T	Oct 22	Xinyu Wang Feedback by Sumit Tarafder	Paper Presentation I [Slides]	Zhang et al. US-align: universal structure alignments of proteins, nucleic acids, and macromolecular complexes. Nature Methods 2022.
9 / R	Oct 24	Luis Lazcano Feedback by Stephen Owesney	Paper Presentation I [Slides]	Mirdita et al. ColabFold: making protein folding accessible to all. Nature Methods 2022.
10 / T	Oct 29	Sumit Tarafder Feedback by Xinyu Wang	Paper Presentation I [Slides]	Jing et al. Fast and effective protein model refinement using deep graph neural networks. Nature Computational Science 2021.
10 / R	Oct 31	No Class		Midterm Project Proposal Due
11 / T	Nov 5	Stephen Owesney Feedback by Luke Elder	Paper Presentation II [Slides]	Liu et al. PLMSearch: Protein language model powers accurate and fast sequence search for remote homology. Nature Communications 2024.
11 / R	Nov 7	Luke Elder Feedback by Luis Lazcano	Paper Presentation II [Slides]	Feng et al. Integrated structure prediction of protein-protein docking with experimental restraints using ColabDock. Nature Machine Intelligence 2024.
12 / T	Nov 12	Sumit Tarafder Feedback by Xinyu Wang	Paper Presentation II [Slides]	Jing et al. Single-sequence protein structure prediction by integrating protein language models. PNAS 2024.
12 / R	Nov 14	Xinyu Wang Feedback by Sumit Tarafder	Paper Presentation II [Slides]	Jones and Thornton.The impact of AlphaFold2 one year on. Nature Methods 2022.
13 / T	Nov 19	Stephen Owesney Feedback by Luke Elder	Paper Presentation II [Slides]	Schneider et al. When will RNA get its AlphaFold moment? Nucleic Acids Research 2023.
13 / R	Nov 21	No Class
Thanksgiving Break
14 / T	Dec 3	No Class
14 / R	Dec 5	Luis Lazcano Luke Elder Stephen Owesney	Project Presentations (20 minutes each)
15 / T	Dec 10	Sumit Tarafder Xinyu Wang	Project Presentations (20 minutes each)	Project Final Report Due

Assignments

Homeworks
Students are expected to work individually on 3 HWs over the course of the semester that are intended to check your understanding of the lectures and related readings. HWs will written assignments to be submitted electronically via Canvas.

Paper Presentations
Every student will be presenting 2 papers over the course of the semester by selecting from this list and delivering two in-class presentations for 45 minutes. The goal is to educate the others in the class about the topic, so think about how to best cover the material, do a good job with slides, and be prepared for lots of questions. The topics and scheduling will be decided at the beginning of the semester.

Here is a suggested outline for the presentation:

Introduction and context

Motivation -- Why do we care?
Background -- What is the current state of the art? Challenges? Related methods?

A clear statement of the task or research question posed by the paper
Method

Technical primer
What is the input and output of the model?
What is the data that the model is trained on?

Evaluation / Results
Conclusions
Future directions
Concluding questions or reflections for discussion

The paper presenters should upload the presentation slides to Canvas no later than 24 hours before their paper presentations.

Post-lecture feedback: Each week, 1 non-presenter will be in charge of providing feedback for the presenters. A few bullet points of constructive feedback on what they liked, what was clear, and suggestions for improvement will be due by the end of the day on Friday of a given week. During and after the presentation, all students should be prepared to discuss the papers together in a round-table format.

Midterm Project Proposal
In this class, a key deliverable will be to compose a research proposal that is due on October 31 following the guidelines of the National Science Foundation (NSF) Graduate Research Fellowship Program (GRFP). There are a number of benefits of this exercise, including:

practice evaluating the current research literature and thinking creatively about potential research directions for thesis research
practice writing a research proposal, an exercise that will become increasingly important in your research career
if you are an eligible 1st or 2nd year graduate student or thinking about applying to Ph.D. programs this year, the deadline of this assignment is such that it should be possible to submit the proposal as part of the full NSF GRFP application (deadlines in mid-October)

Receiving the NSF GRFP will fund your Ph.D. research, which can give flexibility on what you work on, and is a very nice addition to your CV. For this assignment, the task will be to write the research proposal, following the guidelines for the research plan component of the NSF GRFP application. You are free to format the subsections of your assignment as you see fit. However, the overall formatting of the essay should follow the NSF instructions exactly. The proposal should be structured and written as if it were being submitted, i.e. it should clearly communicate the proposal's Intellectual Merit and Broader Impacts (see the program solicitation), the two review criteria for the NSF GRFP.

NSF GRFP Program Solicitation

The statements must be written using the following guidelines (see the program solicitation):

standard 8.5" x 11" page size 11 point or higher font, except text that is part of an image
Times New Roman font for all text, Cambria Math font for equations, Symbol font for non-alphabetic characters (it is recommended that equations and symbols be inserted as an image)
1" margins on all sides, no text inside 1" margins (no header, footer, name, or page number)
No less than single-spacing (approximately 6 lines per inch) Do not use line spacing options such as “exactly 11 point,” that are less than single spaced PDF file format only
The maximum length of the Graduate Research Plan Statement is two (2) pages (PDF). These page limits include all references, citations, charts, figures, images, and lists of publications and presentations.

Additional Advice

Caveat: This advice is geard towards hypothesis-driven, applications-oriented research, which may be less relevant for pure theory. Make sure your hypothesis is clear. There should be a sentence in your proposal (italicized) beginning "It is hypothesized" ... or similar. Make sure it is stated in the form of a hypothesis, not an aim. For this purpose, a reasonable dictionary definition of the word hypothesis is: A tentative explanation for an observation, phenomenon, or scientific problem that can be tested by further experimentation.
Make sure your aims are appropriately specific.
An aim should describe a specific piece of knowledge you hope to determine, not necessarily the technique used.
Be explicit about what you will do. Describe what you are trying to test, what datasets, and what evaluation methods you will use or design.
Make sure the scope of the proposal is appropriate (not too much or too little) for a Ph.D. thesis (i.e. ~3-4 years of research by 1 graduate student). If in doubt, consult with senior graduate students, postdocs, or faculty.
You are encouraged to include a figure.
Make sure all acronyms are defined. Different acronyms are standard in different communities, and the hypothetical reviewer may not be familiar with your proposed research area.
The intro should be about half a page, not more.
Add a paragraph labeled "Intellectual Merit" at the end.
Add a paragraph labeled "Broader Impacts" at the end.
State anticipated results (particularly exciting results) and how you would interpret them and what follow up experiments or analyses you would do, if any.
It is often times more challenging to concisely communicate a research topic, so treat the 2 page limit as a challenge.
You can use numbered references to save space.

More information on applying for the NSF GRFP:

https://www.alexhunterlang.com/nsf-fellowship
https://mitcommlab.mit.edu/broad/commkit/nsf-research-proposal/

Final Project Report
The final project report, due on December 10, should describe the project outcomes in a self-contained manner, structured around producing a workshop paper that would appear at a machine learning conference, e.g. MLSB at NeurIPS. The goal of this project is to help you think creatively and critically about possible research directions in the field and get hands-on experience. The project must be done individually in this semester (i.e., no double counting). Your final project report is required to be between 7 - 8 pages by using the NeurIPS style files, to be submitted electronically via Canvas. Please use this template so we can fairly judge all student projects without worrying about altered font sizes, margins, etc. The submitted PDF may link to supplementary materials including but not limited to code, open access software package via GitHub, project webpage, videos, and other supplementary material.

Participation
Note that a major part of the grade is attending class each week and engaging in discussions. It is important to monitor your own participation. Participation in discussions includes posing and answering questions, explaining background material or literature, providing constructive feedback on papers, and brainstorming future avenues of research. Attendance is required in all classes. If possible, the instructor must be notified in advance of any extenuating circumstances that lead to missed attendance.

Policies

Late policy for deliverables
Late submissions are NOT allowed (i.e., NOT graded).

Regrading requests
Requests for regrading due to grading errors must be submitted via email within one week of the release of grades.

Services for Students with Disabilities (SSD) accomondation
Any student who has been confirmed by the University as having special needs for learning must notify the instructor during the first week of classes. Such students are encouraged to work with The Office of Services for Students with Disabilities to help coordinate accessibility arrangements.

Academic integrity
The Undergraduate Honor Code pledge that each member of the university community agrees to abide by states: "As a Hokie, I will conduct myself with honor and integrity at all times. I will not lie, cheat, or steal, nor will I accept the actions of those who do."

Students enrolled in this course are responsible for abiding by the Honor Code. A student who has doubts about how the Honor Code applies to any assignment is responsible for obtaining specific guidance from the course instructor before submitting the assignment for evaluation. Ignorance of the rules does not exclude any member of the University community from the requirements and expectations of the Honor Code. For additional information about the Honor Code, please visit: https://www.honorsystem.vt.edu/

Students enrolled in this course are responsible for abiding by the Honor Code. A student who has doubts about how the Honor Code applies to any assignment is responsible for obtaining specific guidance from the course instructor before submitting the assignment for evaluation. Students are strongly encouraged to consult their faculty members regarding the use of any outside materials as the misuse of these sources may constitute a violation of the Honor Code. Ignorance of the rules does not exclude any member of the University community from the requirements and expectations of the Honor Code.

All assignments submitted shall be considered “graded work” and all aspects of your coursework are covered by the Honor Code. All projects and homework assignments are to be completed individually in this course unless otherwise specified. All written work must be written without help from other sources or people, except for the course instructor, the course TAs, and Student Success Center tutors. It is a violation of the Honor Code in this course to receive help from any other source, including online tutoring sites (including but not limited to Chegg, CourseHero, or GroupMe), or generative AI tools (including but not limited to ChatGPT, GitHub Copilot, and Microsoft Copilot).

The Academic Integrity expectations for Hokies are the same in an online class as they are in an in-person class. Hokies are expected to meet the academic integrity standards at Virginia Tech at all times.

Commission of any of the following acts shall constitute academic misconduct. This listing is not, however, exclusive of other acts that may reasonably be said to constitute academic misconduct. Clarification is provided for each definition with some examples of prohibited behaviors in the Undergraduate Honor Code Manual:

Cheating: Cheating includes the intentional use of unauthorized materials, information, notes, study aids or other devices or materials in any academic exercise, or attempts thereof.
Plagiarism: Plagiarism includes the copying of the language, structure, programming, computer code, ideas, and/or thoughts of another and passing off the same as one's own original work, or attempts thereof.
Falsification: Falsification includes the statement of any untruth, either verbally or in writing, with respect to any element of one's academic work, or attempts thereof.
Fabrication: Fabrication includes making up data and results, and recording or reporting them, or submitting fabricated documents, or attempts thereof.
Multiple Submission: Multiple submission involves the submission for credit – without authorization from the instructor receiving the work – of substantial portions of any work (including oral reports) previously submitted for credit at any academic institution of attempts thereof.
Complicity: Complicity includes intentionally helping another to engage in an act of academic misconduct, or attempts thereof.
Violation of University, College, Departmental, Program, Course, or Faculty Rules: The violation of any University, College, Departmental, Program, Course, or Faculty Rules relating to academic matters that may lead to an unfair academic advantage by the student violating the rule(s).

Note that all electronic work submitted for this course is archived and subjected to automatic plagiarism detection and cheating analysis.
If you have questions or are unclear about what constitutes academic misconduct on an assignment, please speak with the instructor. We take the Honor Code seriously in this course. The normal sanction we will recommend for a violation of the Honor Code is an F* sanction as your final course grade. The F represents failure in the course. The identifies “*” a student who has failed to uphold the values of academic integrity at Virginia Tech. A student who receives a sanction of F* as their final course grade shall have it documented on their transcript with the notation “FAILURE DUE TO ACADEMIC HONOR CODE VIOLATION.” You would be required to complete an education program administered by the Honor System in order to have the “*” and notation “FAILURE DUE TO ACADEMIC HONOR CODE VIOLATION” removed from your transcript. The “F” however would be permanently on your transcript.

Principles of community
Because the course will include in-class discussions, we will adhere to Virginia Tech Principles of Community.

Student well-being support
Supporting the mental health and well-being of students in our class is of high priority to us and Virginia Tech. If you are feeling overwhelmed academically, having trouble functioning, or are worried about a friend, please reach out to any of the following offices:

Cook Counseling:
- 540-231-6557 to schedule an appointment and/or 24/7 crisis support
- ucc.vt.edu for more information
Dean of Students Office:
- 540-231-3787 for general advice
- 540-231-6411 for after-hours crisis
- dos.vt.edu for more information
Hokie Wellness:
- hokiewellness.vt.edu for more information about health and wellness workshops and consultations
Services for Students with Disabilities

540-231-3788 or ssd.vt.edu for more information about accommodations and other disability-related supports

Student Success Center:
- The Student Success Center helps students develop the skills needed to accomplish their academic goals and become self-directed learners. Their free services include individual and group tutoring, peer academic coaching, a Seminar Series on Academic Success, and more. Students can book appointments through Navigate. For instructions and more information, please visit www.studentsuccess.vt.edu.

For a full listing of campus resources check out well-being.vt.edu.

Please also feel free to speak with the instructor. We will make an effort to work with you; we care about you.

Technical support
Technical: For technical support assistance regarding any problems with Canvas, Zoom, or e-mail, please contact 4Help. For technical support issues related to Web-CAT or CodeWorkout, send an e-mail request to webcat@vt.edu providing specific details of the problem or issue. For questions related to programming assignments or homework, ask questions on the class general discussion area on Canvas, where one of the instructors or TAs can provide answers.
Canvas privacy policy: http://www.canvaslms.com/policies/intl-privacy.