Data science for infectious disease and cell biology @czbiohub.
Dec 31, 2025 • 9 tweets • 3 min read
Because I somehow can't stop, I made a library for doing interpretability on open weights models that works -- fast -- on your mac laptop. No ssh, no colab. 90 seconds from git clone to logit lens on gemma-2.
A colleague asked about the Lambda variant, which has been dominant in Peru since March.
Cases have been observed in the US. Should we be concerned?
NO.
Reason 1: Cases have been observed in 45 US states and taken off in none.
outbreak.info/situation-repo…
Reason 2: Brazil and Chile, which border Peru, have both seen importations of Lambda (aka C.37).
In both cases, Gamma (P.1) is or has become the dominant lineage, and Delta (B.1.617.2) is rising.
Jun 30, 2020 • 16 tweets • 6 min read
I want to talk about a toy model for reasoning about what viral genomics can and cannot tell us about #SARSCoV2 transmission.
Suppose viral isolates from two people have *identical* genotypes. How many transmission events separate them?
1/13
The answer can vary widely by on chance (mutation is a random process after all).
But, more surprisingly, the range of variation depends a lot on the phase of the outbreak, in a way we can quantify.
2/13
Mar 13, 2020 • 5 tweets • 3 min read
@alexismadrigal Alexis, I just spoke with a few of my colleagues who were on the panel, who said "these inaccurate notes were not reviewed, endorsed" and are missing significant context.
In particular, the large % infected is not a forgone conclusion; coordinated action now can help.
@alexismadrigal In China, Korea, and Singapore, we have seen that 1) this virus is containable, at least in the short term, and 2) that 1) is true in diverse population structures, political systems, using different primary response methods.
Single-Cell RNA-seq data is notoriously noisy. There are dozens of methods for cleaning it, but how do you know when they’re working right?
How clean is too clean?
Presenting...Molecular Cross-Validation
biorxiv.org/content/10.110…
Where does the noise come from? If a cell has 500k molecules and a droplet experiment nets you 5k, then you’re looking at just 1% of what was there. Lots of lowly-expressed genes will be missing by chance (the phenom formerly known as "dropout").
