Twitxy Saved

Geir Kjetil Sandve

@SandveGeir

Professor in bioinformatics/machine learning at University of Oslo, Norway. Follow me at @SandveGeir@mstdn.science or at @SandveGeir@genomic.social
sandvelab.org Joined July 2017
100 Following 780 Followers 227 Posts

Latest posts 5

Geir Kjetil Sandve @SandveGeir
Another way to beware of intuition around statistics in life science - with hundreds of significant findings, it may seem obvious that at least some must be true. But not necessarily. A thread by Chakri based on a great collaboration with Mamica, Emilie, Manuela and Jessica!
1/12 🧵 Quick thread: a heads-up for anyone working with high-dimensional omics data and multiple testing — FDR correction can sometimes lie to you. A thread based on our recent publication in @GenomeBiology : doi.org/10.1186/s13059…
View quoted post on X
Geir Kjetil Sandve @SandveGeir
It took us some time to chew on this, and a bunch of (intriguing) nuances came up as we dug deeper, but I think it landed as a quite accessible and useful overview of concerns and opportunities connected to causality and study design for biomarkers! (with AIRR as case)
Our perspective on the generalization of machine learning-identified biomarkers is now published in @NatMachIntell 🎉 nature.com/articles/s4225…
View quoted post on X
0 replies 5 reposts 6 likes 1.2K views View on X
Geir Kjetil Sandve @SandveGeir
It's not often a new method ends up being both faster and more accurate than existing methods, so I believe KAGE2 by @IvarGrtytten and @knutrand (powered by @BioNumPy) will be a great addition to the toolboxes of structural variant discovery researchers  biorxiv.org/content/10.110…
www.biorxiv.org KAGE 2: Fast and accurate genotyping of structural variation using pangenomes Structural variation is known to play an important and often overlooked role in regulating disease and traits, but accurately detecting structural variants from sequencing data has traditionally been...
KAGE2 is out! Enables very fast and accurate genotyping of structural variants using pangenomes: biorxiv.org/content/10.110…. I’ve spent the last 6+ months going deep into the SV rabbit hole, and had some surprises I thought it’s worth to also share (1/6)
View quoted post on X