A research team synthesized a series of short peptides—including DILT1 through DILT5 and related sequences—using automated solid-phase peptide synthesis to create multichannel nanofibrils. The peptides self-assemble into structures whose lattice geometry is dictated by their specific amino acid sequences, producing hexagonal channel arrangements at the nanoscale.
A study examined how assigning different weights to training sequences affects loss reduction in language models of varying sizes, using both proprietary and open-weight models. It found that the relationship between sequence weight and learning is not linear across scale: small models learn general patterns regardless of weighting, medium-scale models learn patterns roughly proportional to their assigned weights, and large models again learn broadly regardless of weighting.
A research team describes the methodology used to design DNA scaffold and compute domain sequences for a molecular computing system, using scaffolds of varying length (120, 288 and 624 bases) derived partly from biologically sourced M13 DNA. They applied a thermodynamic sequence design approach with strict energy constraints to ensure reliable binding while minimizing unwanted cross-reactions between strands.
R/GA has introduced BRDNA Sequencer, an AI-driven platform designed to help brands build advertising campaigns that adjust in real time based on live data and local conditions. The tool aims to simplify how designers and clients manage AI-assisted workflows while producing more responsive, personalized creative output.
A Reddit user analyzed their extended Spotify streaming history for a 2,000-song playlist and found a 72-track sequence played in the exact same order on two consecutive days. The repetition occurred even with the 'Fewer Repeats' setting enabled and without changing devices or playlists. Other users report similar patterns, including one case where 27 of 30 upcoming tracks matched a prior queue.
Malva is a search system that lets scientists query massive single-cell and spatial transcriptomic datasets using gene names, DNA sequences, or plain-language descriptions like 'cells expressing markers of neurodegeneration.' It combines a language model, a k-mer-based search index, and a metadata service to return annotated cell results in formats compatible with common analysis tools such as scanpy.