Issue 7

April 29, 20265 Articles

Alpha¹ Editorial

This week's top 1% preprint from AI design of proteins, neural circuits and digital twins in mice, to selection of ancient humans

In This Issue

1Generative design of sequence specific DNA binding proteins
2Identification of a neural circuit that enables safe, long-term torpor in mice
3A functional 'digital twin' of retinal input to the mouse superior colliculus
4Protein design for intrinsically disordered proteins using language-based RL (IDiom)
5Local ancestry inference identifies robust evidence of selection in Neolithic Europe

Article 1 of 5Α1View on Alpha¹

Generative design of sequence specific DNA binding proteins

Read Preprint Full Alpha¹ Analysis X Discussion

Summary

Get the top 1% of preprints in your inbox

Create a free Αlpha¹ Science account to follow peer-validated research, receive editorial curations, and participate as a reviewer or author.

ISSUES TERMS PRIVACY FULFILLMENT POLICY

In a protein-design party gone wild, RFdiffusion3 and friends ditch the usual structure-first diffusion drama to whip up custom DNA-binding proteins with hilarious precision, turning abstract sequence dreams into reality like a molecular matchmaker on steroids.

Community Discussion

Posted by Raktim Mitra (@Raktim7879) with a fun GIF, it earned enthusiastic praise as "the most fun paper I have ever written" from the team and broad excitement across the protein design community

Article 2 of 5Α1View on Alpha¹

Identification of a neural circuit that enables safe, long-term torpor in mice

Read Preprint Full Alpha¹ Analysis X Discussion

Summary

Mice get a built-in hibernation button thanks to a newly mapped neural circuit that flips them into safe, long-term torpor mode—basically nature's quirky snooze feature for surviving tough times without the usual risks.

Community Discussion

Highlighted with a big "WOW!!" by Bryan Roth (@zenbrainest), drawing quick interest and shares from neuroscientists intrigued by the circuit discovery

Article 3 of 5Α1View on Alpha¹

A functional 'digital twin' of retinal input to the mouse superior colliculus

Read Preprint Full Alpha¹ Analysis X Discussion

Summary

Neuroscientists built an openly playable "digital twin" of mouse retinal inputs to the superior colliculus, letting you virtually poke neurons with any stimulus (like looming threats) before wasting months in the lab—finally, hypothesis testing with zero mouse guilt.

Community Discussion

Shared by Katrin Franke (@kfrankelab) with detailed thread, Colab link, and team shoutouts, it drew likes and interest from computational/systems neuroscientists for its practical virtual lab potential

Article 4 of 5Α1View on Alpha¹

Protein design for intrinsically disordered proteins using language-based RL (IDiom)

Read Preprint Full Alpha¹ Analysis X Discussion

Summary

While everyone obsesses over rigid protein structures, this team flips the script with language-model RL to design messy intrinsically disordered proteins, complete with a playful web demo that feels like AI mad-libs for molecular chaos.

Community Discussion

Posted by Grant Rotskoff (@grantrotskoff) highlighting the shift from "structure-first" diffusion, it gained solid engagement and bookmarks from the protein design/ML crowd

Article 5 of 5Α1View on Alpha¹

Local ancestry inference identifies robust evidence of selection in Neolithic Europe

Read Preprint Full Alpha¹ Analysis X Discussion

Summary

Ancient DNA sleuths use clever local ancestry tricks to spot strong selection signals in Neolithic Europeans, revealing how our ancestors genetically adapted in quirky, population-specific ways during big farming transitions.

Community Discussion

Shared by Nrken19 (@nrken19) with a clear figure, it resonated with aDNA and anthropology enthusiasts for its robust methods