pbousquets.bsky.social

11 May 2026 16:19

We then thought what else we could learn with PHYFUM. We implemented four different crypt division models in PHYFUM, and found we have signal in simulated data to detect them.

pbousquets.bsky.social

11 May 2026 16:19

We applied PHYFUM to 22 patients (colon, small intestine, endometrium). Gut glands diverged early, with most branching happened during childhood/puberty. Endometrial glands showed much more recent divergence, consistent with cyclical glandular renewal 🔄

pbousquets.bsky.social

11 May 2026 16:19

We carried out simulations under a number of conditions, and found that the median parameter error: 5.6%. Tree reconstruction error (wRF) is also low, indicating that we can successfully reconstruct our simulated phylogenies

pbousquets.bsky.social

11 May 2026 16:19

Building on previous work, we developed a mathematical model using methylation in CpGs that stochastically switch over time (fCpGs). Because this is happening in a large number of sites and rapidly enough, we have resolution to see evolution in normal tissue

pbousquets.bsky.social

11 May 2026 16:19

Studying how normal tissues evolve in living humans is incredibly hard. Most phylogenetic methods rely on DNA mutations, but mutation rates are too low to get fine-grained resolution of tissue history. We needed a faster molecular clock ⏱️