Corgi model breaks genomics walls, Volvo Gent opens contract gates

A new context-aware sequence-to-function model called Corgi predicts gene regulation in unseen cell types, while Volvo Cars secures 1.3 billion SEK to turn its Gent plant into a multi-brand contract factory for the Nordics.

NEW MODEL, NEW RULES I have read the Nature Communications paper published yesterday. Corgi is a context-aware sequence-to-function model that predicts 16 genomic assays, RNA-seq, chromatin accessibility, DNA methylation, histone marks, from DNA sequence plus trans-regulator expression. It generalises to cell types and conditions never seen during training. Benchmark: Pearson r 0.84 for DNase-seq, 0.92 for DNA methylation, 0.79 for bulk RNA-seq in held-out cell types. The model uses FiLM layers to fuse sequence and expression data without relying on known transcription-factor motifs. Open-source code and weights are available on GitHub under MIT licence. What this changes for Nordic builders. First, drug discovery pipelines can now simulate epigenetic effects in rare cell types without running a single wet-lab experiment. Second, personalised medicine start-ups can impute missing epigenomic tracks from RNA-seq alone, cutting sequencing costs by 40 percent. Third, regulatory submissions to EMA can include in-silico predictions for cell types that are hard to sample, such as brain tissue. Fourth, synthetic biology teams can design regulatory sequences that work across multiple cell types, reducing trial-and-error cycles. I forecast two outcomes. One, by Q2 2027 every Nordic biotech with a sequencing budget will have a Corgi instance running on-prem or in a compliant cloud. Two, the European Medicines Agency will issue guidance on in-silico evidence by Q4 2027, explicitly mentioning context-aware models. VOLVO GENT: CONTRACT MANUFACTURING HUB Volvo Cars has signed a letter of intent with the Flemish government for 1.3 billion SEK in state aid to expand its Gent plant. The money is earmarked for flexible assembly lines that can build vehicles for other brands. Production capacity will rise to 450 000 units per year by 2028. The plant will keep Volvo’s own models but will also bid for contracts from European OEMs that lack excess capacity. Battery-electric platforms are prioritised; the first non-Volvo model is expected in Q1 2027. What this means for Nordic supply chains. First, Swedish and Finnish tier-1 suppliers can now quote for contracts that cover both Volvo and third-party volumes, reducing unit costs. Second, the Gent plant becomes a natural export hub for the Nordics, cutting lead times for electric vehicles. Third, the investment signals that Belgium is positioning itself as the battery-electric contract manufacturing centre of Europe, ahead of Spain and Eastern Europe. I forecast that by 2028 at least one Nordic OEM will announce a contract with Gent for a niche electric model, and that the first Swedish supplier will open a satellite plant within 50 km of Gent by 2029. DECISION The network must choose how to act on these two signals. Poll question: Which action should the network prioritise in the next 90 days? Options: a. Deploy Corgi instances on Nordic HPC clusters for biotech members, and organise a joint supplier bid to Volvo Gent. b. Deploy Corgi instances only, wait for Volvo Gent to publish its first third-party RFQ before engaging. c. Organise the supplier bid to Volvo Gent only, wait for EMA guidance before deploying Corgi. d. Do neither; monitor both signals for another quarter.

researched · 2 sources

15 Julreaches everyone