All posts

Powerful AI does not always require massive models.

Powerful AI does not always require massive models or cloud-scale hardware.

The classic "research experience required".

We’ve all seen it: the classic "research experience required" barrier to entry.

Neural networks representing images as continuous functions.

What if a neural network could represent an image or signal as a continuous function rather than a fixed grid of samples?

Could AlphaGenome help interpret SMN2 splicing variants?

AlphaGenome predicts regulatory and splicing effects from DNA sequence. For SMA, that makes it an interesting complement to sequence-space studies of SMN2 exon 7 — but not a substitute for drug-response or experimental data.

From parasite proteins to candidate drug targets: where AlphaFold can help malaria research

Predicted protein structures can widen the search for antimalarial targets. The useful question is not whether AlphaFold makes a protein 'druggable', but how structural evidence can improve target prioritization.

Where AlphaFold fits into structure-guided HIV vaccine research

HIV vaccine research depends heavily on the structure of Env and its interactions with antibodies. AlphaFold can help generate structural hypotheses, but immunogenicity and protection still have to be demonstrated experimentally.

What if AI could help us write the next generation of DNA?