iiinova — where your digital self keeps evolving

Janne Pott @pottj.bsky.social · 23/09/2026

Time off at a nice beach, lots of swimming (fishes, but no jellyfish), and then getting PM that a grant was approved 😁 - what more can you ask for?

020

Janne Pott @pottj.bsky.social · 22/04/2026

Any other ideas for the missing replication? Happy to discuss!

000

Janne Pott @pottj.bsky.social · 22/04/2026

Possible reasons for the lack of replication: 1️⃣ there is truely no causal effect. 2️⃣ there is a causal effect, but working via a mechanism we cannot detect with MR (e.g. functionality vs protein levels). 3️⃣ this is another example where MR misses the signal (like GLP-1R and BMI, or CRP and CAD).

100

Janne Pott @pottj.bsky.social · 06/10/2025

Could this be a survival data problem? Harzard ratio are hazardous (check pmc.ncbi.nlm.nih.gov/articles/PMC... ). Could this be a MR problem (life course effect vs specific effect in cancer patients)? Let me know if you have any thoughts on this!

pmc.ncbi.nlm.nih.gov

The Hazards of Hazard Ratios

000

Janne Pott @pottj.bsky.social · 06/10/2025

First, the reported SNP failed to associate in independent data sets, then the MR with genetically proxied PCSK9 levels was not significant. Finally, we tried MVMR to correct for potential influence from lipids, but still observed no effect.

100

Janne Pott @pottj.bsky.social · 06/10/2025

📣Preprint alert - a tricky one, as there are no significant results (and we all know null-results are hard to publish) Do PCSK9 levels affect survival in breast cancer patients? We tried to replicate findings in cell lines and mouse model using 3 Mendelian Randomization approaches, but failed.

185

Janne Pott @pottj.bsky.social · 10/06/2025

We also compare fixing taxon traceability to the gold standard, phylogenetic decisiveness. Thanks to my master thesis supervisor Mareike Fischer for continuing on this project, keeping me in on it, and especially for proving the bounds!

010

Janne Pott @pottj.bsky.social · 10/06/2025

Fixing taxon traceability (simplified): assume n taxa and k sets of quadruples defining a tree each. Then the set of k quadruples is fixing taxon traceable if all missing quadruples have a fixing taxon or can be resolved using the input quadruples + those with fixing taxa.

100

Janne Pott @pottj.bsky.social · 10/06/2025

Fixing taxon: assume 5 taxa and only 4 of 5 possible trees available. Then one taxon is in all trees (e.g. taxon 3 in the quartet trees on the left, combination of Figure 1, 2, and 4 of the publication), the supertree is already defined (tree on the right), and hence the "missing" tree too!

100