About
Hi! My name is Alaina, and I'm a Canadian data scientist who taught herself molecular and computational biology starting in 2011. I spend a lot of my free time doing bioinformatics—population genetics, ancient DNA, that sort of thing.
This is where I share what I'm working on and the tools I build along the way. The "X" in xbio doesn't mean anything specific. (Exponential biology? The X chromosome? The unknown variable? Pick your favourite.)
I work in AI and health science, but the stuff on this site is just me chasing haplogroups through time because I think it's interesting. And also because I think that I can bring 35+ years of software engineering to make science more reproducible and accessible.
We're a bike-only family living in Toronto, so I'm also very concerned with cycling safety. That's where the Toronto Cycling Data project comes from.
Explorations
Write-ups of what I'm working on. Mostly population genetics and ancient DNA, but I also spend a lot of time building and supervising AI coding agents, so occasionally that shows up here too.
I Supervised AI Coding Agents for Hundreds of Hours. Here's What Goes Wrong.
I analysed 1.65 GB of production traces from Claude Code and Cursor to understand how AI coding agents systematically optimize for "done" over "correct." Seven categories of misalignment, a working detection pipeline, and some worrying findings about what monitors can and can't catch.
No, the Jakobsson Paper Doesn't Disprove Out of Africa
A response to viral misinterpretations of recent African origins research. The paper supports African origins; it just refines the geography.
Exploring the Growth of mtDNA Diversity
Can we reliably identify new haplogroups with constant data input? Building a system to discover missing branches in the mitochondrial tree.
Making Haplogroup Callers Work with Ancient DNA
Ancient DNA is a mess—damage, low coverage, missing data. Here's how I think about building classifiers that can handle it.
Haplogroup Classification at Scale
How yallHap works, and what I learned building a Y-chromosome haplogroup caller that doesn't choke on 185K SNPs or ancient DNA.
Building Your Own Bioinformatics Data Library
How I assembled ~30TB of reference genomes, population panels, and ancient DNA samples to learn population genetics by doing.
Detecting Ghost Populations with f-Statistics
How to find ancestry from populations we've never sampled, using patterns of shared genetic drift.
Tools
Open source tools for genomics research. First-class support for ancient DNA, but tastes great with modern DNA too.
eveHap
Mitochondrial haplogroup classifier
An ancient-DNA-friendly mtDNA haplogroup caller using probabilistic approaches optimized for damaged and low-coverage samples.