Something is brewing at the intersection of
What if we could predict how targeted genetic changes affect living organisms — before ever touching a cell?
What if identifying the genetic markers behind rare diseases didn't require decades of manual research?
What if eigenvalue decomposition and PCA could find meaningful patterns hidden in billions of base pairs?
What if we could virtually predict phenotypic outcomes from genetic modifications — adding a safety layer before the wet lab?
What if deep learning could engineer the features itself — no manual feature selection, no genetic expert bottleneck?
The Idea
We're exploring the space between computational genomics and therapeutic discovery. The premise is simple: use dimensionality reduction and deep learning on large-scale genome sequencing data to predict how targeted genetic modifications might change phenotypic outcomes.
Think of it as virtual gain-of-function research — screening hypotheses computationally, then handing only the most promising and safe ones to biotech researchers for real-world validation.
We're still in the earliest stages. Nothing is built yet. But the question feels worth asking.
Imagine
A glimpse of what PCA on genome data might look like — each dot a sample, each cluster a phenotypic grouping waiting to be understood.
Interested?
We're looking for co-founders, genomics researchers, biotech collaborators, and early believers. If this resonates — let's talk.
No spam. Just a ping when there's something real to share.