What's New in Pipette: From Drug Design to Long Reads

Pipette.bio Team

Apr. 11, 2026

The last six weeks have been about reach. Pipette can now handle entire categories of analysis it couldn't touch before (drug discovery, long-read sequencing, large-scale single-cell, synthetic biology), and the existing skills got smarter about the questions you actually ask.

A Whole New Frontier: Drug Design

This is the biggest addition since we launched. Pipette now has a dedicated drug design queue with its own worker, its own environment, and 12 new skills built around small-molecule discovery and protein engineering.

You can now run molecular docking, virtual screening, ADMET prediction, pharmacophore modeling, and structure-based design, all from the same chat interface you use for RNA-seq. Submit a target protein and a compound library, and the agent will dock, score, filter, and report. No separate tools to install, no environment to manage. The drug design queue runs on its own container so it doesn't compete with your sequencing jobs.

If you've been waiting for Pipette to cross from analysis into discovery, this is it.

Single-Cell at Scale: STARsolo on a Bigger Box

Single-cell FASTQ processing has always been a memory hog. Cell Ranger and STARsolo can chew through 60+ GB on a typical 10x experiment, which is more than our default workers carry.

We've added a new sclarge-q worker (16 vCPU, 60 GB RAM) and a dedicated STARsolo skill that runs the full FASTQ → count matrix pipeline for scRNA-seq. The skill handles barcode whitelist resolution, GTF lookup for human and mouse, and produces a Scanpy-ready output. If you've been uploading pre-processed .h5ad files because raw FASTQs were too big to handle, you can now start one step earlier.

Long Reads, Finally: Nanopore Analysis

Nanopore data has been a gap in our coverage. Not anymore. The new nanopore-analysis skill handles basecalling QC, alignment with minimap2, structural variant calling with Sniffles, and methylation calling. Whether you're running an Oxford Nanopore experiment for SVs, isoforms, or modified bases, the agent now knows what to do with the FAST5 and FASTQ files you give it.

Synthetic Biology: Oligo Design and Golden Gate

Two new skills round out our wet-lab support:

• oligo-design: primer design for PCR, qPCR, sequencing, and cloning, with melting temperature checks, secondary structure scoring, and BLAST specificity validation against a chosen genome.
• golden-gate-assembly: fragment design for Golden Gate cloning, with BsaI/BsmBI site checking, junction optimization, and assembly verification.

These pair naturally with the existing CRISPR and variant skills. You can design a knockout construct, validate the guide RNAs, and check assembly junctions in a single chat session.

Open Chromatin: ATAC-seq

The new atacseq skill handles the full ATAC-seq pipeline: adapter trimming, alignment with Bowtie2, duplicate removal, peak calling with MACS2, and fragment-size QC. It works with both bulk ATAC and the lighter Omni-ATAC protocol, and produces peaks in standard BED format ready for downstream motif analysis or differential accessibility.

Query Skills: COSMIC, STRING, and KEGG Without the Hassle

Three new zero-install query skills let you pull annotation data without spinning up a heavy environment:

• COSMIC: query the Catalogue of Somatic Mutations in Cancer for mutation frequencies, drug sensitivities, and tissue distributions.
• STRING: fetch protein-protein interaction networks for any gene list, with confidence scores and pathway context.
• KEGG: pull pathway maps, gene-pathway memberships, and orthology assignments for any organism.

These run in seconds and cost almost nothing in credits. Use them mid-conversation to enrich a result or sanity-check a finding without re-running an analysis.

CRISPR Off-Targets: Cas-OFFinder

For everyone running CRISPR screens or designing guide RNAs, cas-offinder is now part of the default environment. Provide a guide sequence and a genome, and it returns every potential off-target site with mismatch counts and PAM context, much faster than BLAST-based approaches and aware of CRISPR-specific scoring.

Smarter Single-Cell: Compositional Analysis and Pancreas Annotations

Single-cell skills got two important upgrades:

• compositional-analysis: a new skill for differential abundance testing across conditions. Includes scCODA, milo, and DA-seq, with cross-references from both the Scanpy and Seurat skills so the agent knows when to suggest it.
• Pancreas reference dictionary: both Scanpy and Seurat skills now ship with a curated marker dictionary for pancreatic cell types (alpha, beta, delta, gamma, ductal, acinar, stellate, immune). If you upload a pancreas dataset, annotation just works.

More Organisms in Pathway Enrichment

The pathway enrichment skill now supports a much broader range of organisms, including Brassica napus (oilseed rape) and additional model and crop species. If you're working outside the human/mouse/rat default, your enrichment results will now be biologically meaningful instead of empty.

What This Adds Up To

Pipette now spans the full arc of modern molecular biology, from raw nanopore reads, through bulk and single-cell sequencing, into pathway and network biology, and now all the way out to drug discovery and synthetic construct design. The same chat interface, the same credit system, the same methodology review on every result.

If there's an analysis you've been doing manually because Pipette didn't support it, try it again. There's a good chance it does now.

Try it at pipette.bio. Your first 20 credits each month are free.