We scan new podcasts and send you the top 5 insights daily.
Because DNA encoded libraries are built combinatorially from building blocks, millions of molecules differ by only one or two changes. This structured variation enables machine learning models to isolate exactly which chemical fragments drive target binding while keeping the rest of the molecule constant.
The relationship between a multi-specific antibody's design and its function is often non-intuitive. LabGenius's ML platform excels by exploring this complex "fitness landscape" without human bias, identifying high-performing molecules that a rational designer would deem too unconventional or "crazy."
Instead of digital screening, Sethera Therapeutics generates vast physical libraries of hundreds of trillions of unique peptides within a small test tube. Each peptide is given a unique barcode, enabling researchers to test them against real biological targets and identify effective molecules from infinitesimally small quantities.
Unlike traditional methods that simulate physical interactions like a key in a lock, ProPhet's AI learns the fundamental patterns governing why certain molecules and proteins interact. This allows for prediction without needing slow, expensive, and often impossible physical or computational simulations.
Because traditional wet-lab techniques test only ~50 sequences weekly, Early instituted massively parallel reporter assays testing 250,000 barcoded sequences in a single batch. By funneling high-quality outlier data into DNA-trained large language models paired with predictive oracle filters, each laboratory-to-model cycle compounds predictive accuracy. This flywheel effectively treats DNA as a programmable language, mirroring how AlphaFold unlocked programmable proteins.
Traditional drug discovery groups run DNA encoded libraries against isolated, difficult targets in hope of a single hit, missing systematic library quirks. Screening every library across hundreds of targets surfaces batch-to-batch manufacturing variations, chemical motifs, and assay failure modes that single screens never detect.
Traditional lab-in-the-loop systems require slow robotic synthesis to test model hypotheses. By generating millions of physical binding interactions upfront, language models acting as medicinal chemists can hypothesize, select test compounds, and receive instant grounded feedback as if operating in a coding sandbox.
The next inflection point will come from clever data generation strategies optimized for AI models, not human analysis. This "black box data" approach—like pooled screening with sequencing readouts—is vastly more scalable and creates a powerful, proprietary moat for companies.
Many therapeutic target proteins naturally bind DNA, producing massive false positive readouts in DNA encoded library screens. Assigning dozens of distinct DNA barcodes to identical small molecules reveals whether target enrichment is driven by specific DNA nucleotide motifs rather than the attached chemical.
In a 2,000-team Kaggle challenge using a 4-billion interaction dataset, models performed adequately when test molecules shared building blocks or core hubs with the training set. However, when evaluated against completely orthogonal chemical libraries with distinct chemistry, no model beat random chance.
In multi-specific antibody design, small structural modifications—like altering a linker length or binder position—can cause large, unpredictable shifts in potency, selectivity, and safety. This extreme sensitivity makes traditional, intuition-led engineering unreliable and necessitates data-intensive approaches.