Skip to content

Topic

AI in biology

Use of machine learning in biological research, including predicting protein structures and assigning function to uncharacterised genes.

Current stories

productConfirmed2 publishers

Colossal's GenoMorph borrows from more than 70 bird species to pick about 2,300 dodo edit targets

Colossal Biosciences says its GenoMorph platform has found about 2,300 candidate dodo edit targets while working from a single dodo genome. CEO Ben Lamm says the targets still need lab testing, and the platform edits no genes itself.

Perspective Coverage

4 publishers
Builder
Builder 45%
Operator
Operator 25%
Investor
Investor 30%

Reality

Evidence35
Adoption
Not scored
Hype gap+25
Incentives70
Confidence60
scienceOne report1 publisher

Newly found small-genome viruses suggest encapsulins and viruses evolved from each other repeatedly

University of Miami researchers identified previously unknown viruses that bridge viruses and encapsulin compartments in cells, Phys.org reported. The Nature Communications paper reports evidence that the two lineages evolved from one another repeatedly over millions of years. That contests the theory that encapsulins simply descended from viruses.

Publishers:phys.org

Reality

Evidence55
Adoption
Not scored
Hype gap+15
Incentives
Not scored
Confidence45
investOne report1 publisher

Colossal Biosciences shortlists about 2,300 dodo beak targets with its dodo genome 95% filled

Colossal Biosciences says its GenoMorph AI shortlisted about 2,300 candidate targets for the dodo's beak and that its dodo genome is 95% filled. The company announced no timeframe for hatching a dodo, and the shortlist is still being validated in the lab.

Reality

Evidence35
Adoption
Not scored
Hype gap+30
Incentives60
Confidence50
scienceOne report1 publisher

NTU team mines PubMed with language models to build a 5,118-sequence phase-separation database

NTU Singapore researchers built PLATO, a database of 5,118 protein sequences, and PhaseHub, a prediction platform covering 1,106 species. About a quarter of the entries are proteins that do not condense, a class of results NTU says existing collections underrepresent.

Publishers:phys.org

Reality

Evidence50
Adoption
Not scored
Hype gap+15
Incentives
Not scored
Confidence55
scienceOne report1 publisher

A shape-first search of 214 million predicted proteins turned up a GPCR-family protein inside the cell

Sylvester researchers compared the shapes of more than 214 million predicted proteins, then followed one hit, TM184C, to vesicles riding microtubules into bridges between cells. The human protein rescued a yeast mutant.

Publishers:phys.org

Reality

Evidence62
Adoption
Not scored
Hype gap+15
Incentives70
Confidence55