$TWST $GOOGL Google DeepMind/Isomorphic Labs announced SynthID Bio, a watermarking framework designed to embed verifiable digital signatures into AI-generated protein sequences and biomolecular structures. The synthetic gene fragments encoding the designed protein binders for in vitro validation were codon-optimized and purchased from @TwistBioscience. @adaptyvbio is credited for contributing in vitro validation. James Diggans, Twist's VP of Policy & Biosecurity, served as an expert industry peer-reviewer and provided early feedback on the research from a biosecurity perspective. Standard biosecurity screening algorithms compare incoming DNA orders against databases of known pathogens and toxins based on sequence identity. However, generative AI models can invent completely novel protein sequences that share little to no sequence resemblance to known wild-type organisms, creating screening ambiguities that traditionally require time-consuming manual reviews. SynthID Bio provides Twist and other manufacturers' biosecurity screening pipelines with an automated "trusted origin" verification signal and serves as a critical component in a layered biosecurity model: • Proof of Safe Origin: An intact watermark proves that an unfamiliar or highly novel sequence originated from an AI design model equipped with built-in safety guardrails. • Operational Efficiency: Recognizing a verified watermark allows suppiers to fast-track compliant customer orders, avoid unnecessary screening delays, and potentially pass some cost savings on to customers. • Model-Level Mitigations: AI developers embed safety filters and watermarks at the software generation stage. • Physical Synthesis Gatekeeping: DNA manufacturers like Twist screen orders and verify customer credentials before printing physical DNA. • Traceability & Database Integrity: Watermarks ensure sequence provenance is preserved even if digital metadata is stripped or falsified, protecting public repositories (like GenBank, UniProt, and the Protein Data Bank) from mislabeled synthetic entries. SynthID Bio enables biopharma researchers to scale AI-driven protein discovery while giving synthesis providers like Twist the automated tools needed to maintain biosecurity compliance without stalling commercial turnaround times.

$TWST Twist is becoming a foundational architect of synthetic biology biosecurity. They're a key participant in the Sequence Biosecurity Risk Consortium (SBRC) collaborating with organizations like MIT to stress-test screening algorithms and establish the definitive "Biosecurity Flag Rubric" for the entire industry.

$ABCL Still reeling from AbCellera's phase 2 data and looking forward to next steps for ABCL635. We have a few more milestones coming: 1) Unblinding of the Phase 1 MAD Cohort (Expected late 2026) Small sample size, but I'll be looking for the release of phase 1 redosing data. The Ph 1 trial included a parallel Multiple Ascending Dose (MAD) segment where 16 postmenopausal women received consecutive monthly injections of either 300 or 600mg for 3 months. This data has been kept strictly blinded while patients complete a mandatory 20-week safety follow-up. Management explicitly noted they would not unblind this data until after the Phase 2 top-line readout occurred. 2) 12-Week Follow-up Data (Expected Q4 2026) This data isn't about proving initial efficacy (the primary 4-week endpoint already established that). Because Phase 2 participants only got a single 600mg dose, observing them for an additional eight weeks allows researchers to track exactly how therapeutic relief changes as the drug naturally clears their system. This generates a real-world dose-response curve and builds a robust PK/PD model. I'll be looking for a few things in the 12-week data: • The Rate of Diminishing Efficacy: How quickly do hot flashes return as the antibody washes out? Management fully expects the effect to diminish by week 12, and tracking this decline will pinpoint the exact threshold of drug concentration required to keep the brain's thermostat calm. • Optimal Phase 3 Dose Selection: I'll be looking for confirmation of the 600mg loading dose followed by a 300mg steady-state monthly maintenance dose strategy previously mentioned by management. • Extended Dosing Potential: If the data shows patients maintain strong relief well beyond the four-week mark despite falling drug levels in the blood, it could prompt AbCellera to consider an even more convenient dosing regimen like every 6 or 8 weeks. 3) Open-Label Extension (OLE) Data (Expected Q1 2027) Following the 12-week monitoring period, all Phase 2 participants (including those who were previously in the placebo group) have the option to enter the OLE and receive a 600 mg dose. The OLE is designed to provide long-term safety, tolerability, and durability data while allowing trial participants continued access. Beyond these 3 milestones, I'll also be looking for meetings with the FDA and early details on Phase 3 trial design, as well as the expansion into oncology-related VMS. Exciting times, long and strong.