Issue 2
From xenobots with memories, world model for cells, to RNA-Seq analyses speedup using GPUs
Alpha¹ Editorial
Behavioral, Physiological, and Transcriptional Mechanisms of Memory in a Synthetic Living Construct
Imagine a blob of frog skin with no brain that can still "remember" things. These "xenobots" use weird calcium waves to store memories of chemical hits for days. It’s basically a biological hard drive made of lab-grown slime—proving you don't actually need a brain to have a memory.
Sparked massive hype among synthetic biologists and roboticists, posted by Michael Levin (@drmichaellevin) with 21 replies debating non-neural cognition
Towards building a World Model to simulate perturbation-induced cellular dynamics by AlphaCell
Think of this as "The Sims" but for real biology. AlphaCell treats the inside of a cell like a video game physics engine. It can predict exactly how a cell will freak out when you hit it with a drug or a mutation—allowing scientists to run experiments in a digital world without ever touching a pipette.
Went viral in AI + single-cell circles for its sci-fi “world model” vibes, posted by @razoralign
RNA-seq analysis in seconds using GPUs
Bioinformatics just got a nitro boost. An Icelandic team rebuilt the industry-standard software to run on high-powered GPUs. Now, a massive data crunch that used to take a 40-minute lunch break is finished in 50 seconds. It’s "blink-and-it’s-done" wizardry for genomic data.
Computational biologists lost their minds over the speed, posted by Keith Robison (@OmicsOmicsBlog)
An Energy Landscape Approach to Miniaturizing Enzymes using Protein Language Model Embeddings
This AI tool acts like a shrink ray for proteins. It takes bulky, complex enzymes and designs "Mini-Me" versions that are tiny but still work perfectly. These pocket-sized enzymes are way easier to handle in the lab and are now available for anyone to download and test.
Protein designers and BAGEL fans went wild sharing the open-sourced mini-sequences, posted by Jakub Lála (@jakublala)
CellSweep: Decontaminating Ambient RNA and Barcode Swapping in Single-Cell Data
Single-cell data is usually full of "background noise"—like trying to hear a conversation in a loud stadium. CellSweep acts like a high-powered vacuum cleaner, sucking up all the digital gunk and ambient "ghost" RNA so that scientists can actually see what the cells are doing.
Single-cell Twitter erupted with relief praising the benchmarks, posted by Lior Pachter (@lpachter)
In vivo multiomic Perturb-seq with enhanced nuclear gRNA capture
Mapping genes in a living brain is a nightmare because the "guide" molecules usually get lost. Researchers engineered these guides to stay "glued" inside the nucleus. This creates a crystal-clear map of how genes shape the brain without wasting time on messy, unreadable data.
Neuro + CRISPR crowd lit up over the in-vivo breakthrough, posted by Xin Jin (@xinjin)
Subscribe to Weekly Curation
Weekly curation of the top 1% most discussed bioRxiv preprints, delivered directly to your inbox.