From Newsgroup: talk.origins

https://medicalxpress.com/news/2026-02-big-human-height-scientists-algorithm.html

https://www.cell.com/cell-genomics/fulltext/S2666-979X(26)00024-8

These researchers claim to have developed a means of analyzing huge
genomic and phenotypic data sets to make phenotypic predictions. They
used over 13 million human whole genome sequences and UK biobank data to
feed into their system and claim to be able to make genetic predictions
from the data. They claim to have used 17 million sequence variants.

Their main analysis was for human height, but they tested 13 other
phenotypes. After training their system they claim a 46% prediction
accuracy for human height. This isn't that impressive, and indicates
that we have a lot to learn about how all the genetic variants interact
with each other and with the environment in order to improve prediction accuracy. They seem to have a ways to go in order to make predictions
from the genetics that the individual posesses.

It is a major accomplishment just dealing with data sets this large. We
had issues dealing with hundreds of thousands of genotypes. Dealing
with 13 million is a big deal.

Ron Okimoto

--- Synchronet 3.21b-Linux NewsLink 1.2