Learn the methods well enough to judge what AI tells you.

Three assays, three answers to the same tissue. Each rung buys you resolution and charges you something for it. Knowing which rung answers your question is most of the skill, and it is the spine this course is built on.

Working scientists across domains, not only genomics. Grad students, postdocs, bench researchers, and core-facility users, with mixed to low coding background. If you have never opened a terminal, the first 45 minutes of Session 1 were written for you.

1. Explain, at an intuition level, how modern AI and LLMs work and where they fail.
2. Use AI assistants responsibly to speed up literature review, coding, analysis, and writing.
3. Read and reason about the core statistics and visualizations behind genomics results.
4. Explain what bulk, single-cell, and spatial each measure, and when to use which.
5. Run each assay's analysis on Foundry Connect and interpret the output critically.
6. Apply the discipline: verify, cite, protect data, log for reproducibility.

Six Fridays, September 18 through October 23, every session 1:00 to 4:00 pm. Sessions 4 through 6 climb the ladder one rung at a time, and each opens by contrasting itself with the rung below.

After each session, the feedback link on that row takes two minutes and records your attendance.

All sessions are in Amphitheater II (S4-102), except October 16, which moves to Amphitheater I (S2-102).

1. Session 1
   AI Foundations and Responsible AI

   How language models work and why they are confidently wrong (Eric Ma), how a working bioinformatician flies an AI coding agent (Tommy Tang), then the rules for our data and what an agent looks like on real infrastructure. Setup is pre-work.

   1.0 Before you arrive1.1 Welcome1.2 How AI works1.3 Pilot or passenger1.4 Flight rulesFeedback

   Fri Sep 18 Amphitheater II
   S4-102
2. Session 2
   Statistics and Visualization You Can Trust

   Tommy Tang on the statistics behind a gene list (p-values, multiple comparisons, FDR, q-values) and the six plot types behind most figures in genomics. The toolkit that lets you catch a wrong answer, from AI or anyone.

   2.0 Get set up2.1 Basic statistics2.2 Six plots2.3 Building with AI2.4 Hands-onFeedback

   Fri Sep 25 Amphitheater II
   S4-102
3. Session 3
   AI as a Research Co-pilot

   Eric Ma on AI as a research copilot, worked through one enzyme where mutational scan data, structure, prediction, and the literature answer one question together. John Damask on AI coding agents: build your own AI-maintained knowledge base of the literature on your own computer, search papers, and send agents off to research in parallel. Claude Code or Codex CLI in the terminal; installs are pre-work.

   3.0 Get set up3.1 Why a knowledge base3.2 Your own knowledge base3.3 Research copilotFeedback

   Fri Oct 2 Amphitheater II
   S4-102
4. Session 4
   Bulk RNA-Seq

   Manuel Garber on why differential expression uses models, in four marimo notebooks. Then an AI agent in VS Code sets up an RNA-seq run on Foundry, and you check it and launch it.

   4.0 Get set up4.1 Why DE uses models4.2 Foundry with an agent4.3 HomeworkFeedback

   Fri Oct 9 Amphitheater II
   S4-102
5. Session 5
   Single-cell RNA-Seq

   Cell types and composition, through the Gellatly vitiligo study and the question of why Tregs fail.

   Fri Oct 16 Amphitheater I
   S2-102Room change
6. Session 6
   Spatial: Xenium and seqFISH

   Tissue circuits in situ, through the Wang photosensitive-skin study and its keratinocyte to fibroblast to myeloid circuit.

   Fri Oct 23 Amphitheater II
   S4-102
7. Capstone
   Your own research problem Optional

   Apply the whole workflow to something you are actually stuck on.

   To be scheduled

The afternoon of September 18, in the order it was taught. Each video also sits on its own session page.

The afternoon of September 25, in the order it was taught. Each video also sits on its own session page.

Each session's plan, slides, and links are published as the course runs. A new session is added every week.

No toy datasets. Each domain session teaches its method on a published study from Manuel Garber's lab, with the real data, on the real pipelines. The study supplies the problem; the session is about the technique that answers it.

Instructors

Manuel Garber, PhD Professor · Garber Lab, UMass Chan Medical School

Computational biologist studying how the genome controls gene expression, from non-coding RNAs to the immune response. PhD in Mathematics from Brandeis University, then the Broad Institute, where his methods helped discover lincRNAs. Came to UMass Chan in 2012 to start his lab and direct the Bioinformatics Core.

Ozkan Aydemir, PhD Co-Director, Bioinformatics Core · Assistant Professor

Co-directs the UMass Chan Bioinformatics Core and oversees its operations. Brings extensive experience in high-throughput sequencing data analysis. Faculty in the Program in Molecular Medicine.

Alper Kucukural, PhD Co-Director, Bioinformatics Core · Associate Professor

Co-directs the Bioinformatics Core. Designs and builds reusable, production-grade bioinformatics pipelines and pipeline generation tools for next-generation sequencing data, supporting researchers across UMass Chan Medical School. Faculty in Genomics and Computational Biology and the Program in Molecular Medicine.

Guest speakers

Eric Ma Senior Principal Data Scientist, Moderna

Leads the Data Science and AI Research team at Moderna, working on Bayesian methods for drug discovery. PhD from MIT Biological Engineering, previously at Novartis Institutes for Biomedical Research. A core developer on NetworkX and PyMC, and the author of the open-source packages pyjanitor and nxviz.

Ming “Tommy” Tang Director of Bioinformatics, AstraZeneca

Fourteen years in computational biology across single-cell genomics, epigenomics, and spatial transcriptomics, previously Director of Computational Biology at Immunitas Therapeutics. Teaches bioinformatics to a following of more than 100,000 through the Chatomics newsletter, blog, and YouTube channel, alongside his widely used RNA-seq, ChIP-seq, and scRNA-seq analysis notes.

Eric Kercher Scientist, Alys Pharmaceuticals

Scientist working where biomedical imaging meets cancer research, with industry experience at Alys Pharmaceuticals and Nira Biosciences. Previously a postdoctoral researcher at UMass Chan Medical School. PhD in Physics from Northeastern University, focused on photomedicine, nanomedicine, and optical imaging of the tumor microenvironment.

John Damask Owner, Amroja LLC

Learner and builder, using AI to shrink the time it takes to bring new ideas into the world. Builds open source and commercial software, and helps life sciences and finance organizations accelerate their AI adoption programs. Previously VP of Digital Science at Flagship Pioneering and Global Head of Cloud at Novartis.