Home

Koes Research Group

Removing barriers to computational drug discovery one bit at a time

This repository contains the collected wisdom of the research group. Hopefully the total amount of wisdom (including uncollected) is much larger, but if I had to summarize:

• Look at your data
  • Never assume your data is what you think it should be. What sanity checks have your run on your data? How do you know it is correct? How is it distributed? Are there missing values or outliers? Have you visually inspected a random sample of data items? What is the correlation structure of the data? If your results are good, are they good for the right reasons? How do you know?
• More pictures
  • No walls of text in presentations.

General logistics

• There is a weekly group meeting where one person presents a paper and a subset of people share recent results
• There are weekly one-on-one meetings
• Everyone should be subscribed to the Koes Group Slack Team
• All communication is over slack. Try to prefer #help over direct messages.
• Don't post pictures of code - post the actual code as a code block so it can be copy/pasted.
• Full-time researchers provide daily status reports over slack via the Alice bot. Following agile stand-up meeting conventions, there are two questions.
  • What did you do yesterday
  • What did you do today (answer should include if there is anything you are blocked on)
• For modeling projects, final results, notebooks, data, and methods descriptions should be uploaded to the shared Modeling folder. Sufficient information and detail should be provided so that another person could take over the project.

Presentations

• Generally everyone will present a paper each semester
  • Focus should be on explaining and contextualizing paper
  • 10-20 minutes as needed (methods papers are expected to take longer)
  • Include thoughts on weaknesses of the paper and possible next steps
  • Typically will avoid critique of presentation, but should still strive for good presentation skills
    • Avoid walls of text
    • Okay to use figures from paper
• Full-time researchers and occasionally undergraduates will give a research talk
  • Present your own research project
  • Accessible to general scientific audience unfamiliar with your research
  • Include motivation and background
  • Should be a polished 15-20 minutes presentation
  • Group will critique presentation
  • Figures should be original (preferred) or properly cited.
• A simple Pitt themed template
• Share your presentation in #general before your talk.

Background Reading

• Molecular Interactions
• Pharmacophore Search
• Docking
• Molecular Dynamics
• Protein-Ligand Scoring
• Deep Learning for Structure Based Drug Discovery