Nextflow Global Cache: compute once, reuse everywhere. Introducing a shared caching layer that reuses tasks across runs, pipelines and projects.🧵Lower compute costs, faster results and less storage, from work already done.

❌ Before: Task hashes were tied to each run, so identical tasks re-ran in every new run or pipeline, duplicating outputs.

✅ With Global Cache: Tasks that have already run anywhere in your workspace are restored, not recomputed. No code changes needed.

Read the @SeqeraLabs blog post to learn more about Nextflow Global Cache: hubs.la/Q04DqQgS0

Simple models rival a billion-parameter neural network in an EGFR binding assay. We gave Nextflow Frontier a task: rank molecules by how likely they are to bind EGFR, a cancer drug target. Frontier proposed models, ran them and used their scores to decide what to try next.🧵

This is one example of what Nextflow Frontier enables. Give it a defined computational task and a way to evaluate answers, and it searches candidate workflows and sometimes finds unexpected results.

Full write-up👇Discover how Nextflow Frontier ranked molecules for a cancer drug target with two small chemistry models, trained and run in seconds on a laptop: hubs.la/Q04Ds8-B0

#Nextflow #NextflowFrontier #DrugDiscovery

Nextflow 25.10 introduced workflow params and outputs. 26.04 brought records and static typing. 26.10 completes the picture: composable, reusable pipelines. 💻 Join @tallphil & Ben Sherman (@SeqeraLabs) on Nov 18 for a live walkthrough: hubs.la/Q04DdNSh0

Nextflow Frontier solves and validates biology’s hardest problems through closed-loop discovery.🔃 Give it a problem and a measure of success, and it turns each hypothesis into a reproducible pipeline, tests thousands of them, and plans the next round. hubs.la/Q04C-R7d0

Through Benchling’s AI Connectors, bench scientists can now integrate @SeqeraLabs Platform into @benchling via Seqera MCP.🎉 You can now launch and track Nextflow pipelines, investigate results, and spin up interactive analysis environments, all without leaving Benchling.

We launched Nextflow Frontier: a recursive, closed-loop discovery system for the unsolved problems in life sciences. 📊On BioML-bench (24 agentic tasks), it scored in the 92nd percentile across all categories, outperforming previous published agentic systems by nearly 23%.

🔬 Explore what’s possible with Nextflow Frontier: hubs.la/Q04CBtcM0

📖 Read the Nextflow Frontier blog post: hubs.la/Q04CBvFL0

🌍Two weeks ago, the community came together Barcelona to move the Nextflow ecosystem forward: old friends reconnecting, new faces joining, and everyone learning, collaborating and building together. #NextflowSummit by @SeqeraLabs

🚀 Highlights included watching people run their very first pipelines and seeing long-time maintainers sit side by side with @nf_core newcomers to make meaningful progress.

You don’t need to be an expert. If you’re keen, there’s always someone ready to help you get started. 🔍 See more from this year’s Summit: hubs.la/Q04CKqBz0

Ask AI a question, and you get an answer. Give it a problem and a score, and you get thousands of tested hypotheses. At #AIDDD today, @EvanFloden
showed what experiment-led AI looks like in biology. hubs.la/Q04CDZ7z0

Nextflow Frontier takes your problem and your measure of success, then:

🧪Turns each hypothesis into a reproducible Nextflow pipeline

♻️ Tests thousands of them, then plans the next round

📊 Reaches the 92nd percentile on BioML-bench (24 tasks) vs the 75th for the best published AI system

🧬 Built a model for @ALLOXbio that beat every published method they tested and is now in production

🔒 Your data never leaves your cloud