Life lessons and scientific insight from methods-, hypothesis-, and data-driven research

359f7070cb587948e7da4e1028f5fc41?s=47 Rayna M Harris
February 26, 2020

Life lessons and scientific insight from methods-, hypothesis-, and data-driven research

A short presentation for "Meet and Analyze Data (MAD) at the University of California Davis, a graduate-student-run organization. I describe a handful of life lessons that I've learned and scientific insights that I've gained during my journey as a scientist.

359f7070cb587948e7da4e1028f5fc41?s=128

Rayna M Harris

February 26, 2020
Tweet

Transcript

  1. Life lessons and scientific insight from methods-, hypothesis-, and dataviz-driven

    research Dr. Rayna M. Harris raynamharris Postdoc, UC Davis February 26, 2020 Meet and Analyze Data, UC Davis
  2. In *omics research, these two approach are often pitted against

    each other. Data-driven research Hypothesis-driven research vs.
  3. What drives your scientific pursuit? Is it hypotheses? data? methods?

    Or is is curiosity? geography? availability? How has it changed over time?
  4. Part 1: Lessons learned in my scientific career Methods- driven

    PhD student Geographically- driven intern Curiosity- driven technician Data- driven Postdoc By Caitlin Friesen By Rayna Harris docktodish.com By Rayna Harris Etsy
  5. Part 2: Custom colors and themes for data-visualization in R

  6. What did I learn as a geographically- driven intern?

  7. Immersion is great for learning languages. Me in Costa Rica

    learning to speak spanish while diving for specimens and culturing microscope mushrooms that live in the ocean.
  8. Learning a new language opens a whole new world of

    opportunity (sometimes years later). Me translating Carpentry lessons into Spanish at a hackathon in Germany and teaching UNIX in Argentina.
  9. As a result, I highly recommend learning multiple programming languages.

  10. What did I learn a as curiosity-driven lab technician?

  11. Lab websites can be a little out of date. What

    I wanted to do every day: SCUBA dive What I actually did every day: hormone assays and qPCR Lab website: We study social behavior in fish from the Great Lakes of Africa
  12. Applying the same methods to diverse questions can lead to

    many publications. Unified by a central hypothesis. Not unified. Collaborative.
  13. What did I learn as a methods-driven PhD student?

  14. My methodology of choice was something I called “reverse genomics”

    at the time. Harris, R. M., & Hofmann, H. A. (2013). Ecological Genomics, 149–168. doi:10.1007/978-94-007-7347-9_8 Subject of my NSF GRFP Subject of my PhD thesis
  15. How do mouse brains store memories? Me, a data-driven explorer:

    Look at the activity of all these immediate early genes, like Arc, Jun, Egr1, and Fos! Hypothesis-driven collaborator: What about genes encoding proteins that stabilize memory, like PKM, CAMKII, KIBRA and NSF? Harris RM, Kao HY, Alarcón JM, Fenton AA Hofmann, HA. bioRxiv 2020.02.05.935759; doi: https://doi.org/10.1101/2020.02.05.935759
  16. Good communication is critical for collaboration for speedy collaboration. Hypothesis-driven

    explorers Data-driven researchers questions results
  17. I learned to use and write workflows. I recommend learning

    to automate them too. https://github.com/raynamharris/ IntegrativeProjectWT2015 Harris RM, Kao HY, Alarcón JM, Fenton AA Hofmann, HA. bioRxiv 2020.02.05.935759; doi: https://doi.org/10.1101/2020.02.05.935759
  18. Protect yourself from sample and data loss! A note that

    kept my samples alive when this freezer died. Lost data is easier to recover when you have a backup.
  19. RNA-seq is useful but insufficient for identifying cell types. Additional

    analyses are needed. Northcutt AJ, et al. 2019 PNAS 2019, 116 (52) 26980-26990; DOI: 10.1073/pnas.1911413116
  20. What am I learning as a data-driven postdoc.

  21. Open and reproducible research requires a lot of infrastructure and

    community support. Data Commons Pilot Phase Consortium
  22. It can be difficult to find meaning quickly in large

    RNA-seq projects. https://github.com//macmanes-lab/ DoveParentsRNAseq
  23. `factoextra` can identify the genes that load most strongly on

    principal components https://github.com//macmanes-lab/ DoveParentsRNAseq
  24. `apaTables`saves time and facilitates reproducible writing reporting ANOVA results as

    data frames. 24 https://github.com/macmanes-lab/DoveParentsRNAseq
  25. `wgcna`is great for finding coexpressed genes https://github.com//macmanes-lab/ DoveParentsRNAseq

  26. `sonify`can transform data into music! https://github.com//raynamharris/musicalgenes

  27. We can learn more by working together and combining different

    approaches. Data-driven explorers Hypothesis-driven researchers Methods-driven developers tools results knowledge questions Imagine inspired by lab meeting led by C. Titus Brown
  28. Many thanks to folks in the Instituto Nacional de Biodiversidad,

    Costa Rica College of Natural Science, UT Austin Neural Systems & Behavior Course, MBL Data-Intensive Biology Lab, UC Davis Birds, Brains & Behavior Lab, UC Davis Science & Technology Studies, UC Davis Funded by NSF, NIH
  29. What questions do you have? Dr. Rayna M. Harris @raynamharris

    Postdoc, UC Davis
  30. And now, some hands-on dataviz making custom themes and colors

    https://github.com/raynamharris/tutorials