From Sequence to Structure

One reproducible workbench for every question between genome and molecule.

Mistfold is a computational biology workbench that brings structure prediction, genomic alignment, molecular docking and single-cell analysis into one reproducible environment. Every run is versioned, every parameter is recorded, and every result can be traced back to the raw data that produced it.

Point it at a FASTA file, a VCF, or a cloud bucket of reads. Mistfold provisions GPUs on demand, chains the right tools together, and hands you publication-ready figures — without a single hand-written cluster script.

0.92median pLDDT

40×faster alignment

12Mcells indexed

100%reproducible runs

Start a free project Explore modules

the instruments

Modules of the Workbench

Four instruments, one shared data model. Outputs from one module feed directly into the next — a predicted fold becomes a docking target, a variant call becomes a structural hypothesis.

Fold compute intensity high: Predict 3D protein structures from sequence, including multimers and complexes. Confidence scores are mapped residue-by-residue so you know exactly which loops to trust.

Monomer & multimer prediction
Per-residue confidence overlays
Batch runs across whole proteomes

built onAttention-based folding networks

outputsPDB · mmCIF · PAE maps

Align compute intensity medium: GPU-accelerated short- and long-read alignment with variant calling built in. Go from raw reads to annotated variants in a single pipeline that scales from one sample to a biobank.

Illumina, ONT & PacBio reads
SNV, indel & structural variant calls
Pangenome reference graphs

built onSeed-and-extend · graph alignment

outputsBAM · CRAM · VCF

Dock compute intensity high: Screen ligand libraries against predicted or experimental structures. Pocket detection, pose scoring and free-energy refinement run as one ranked campaign.

Automatic binding-pocket detection
Million-compound virtual screens
Pose clustering & interaction maps

built onDiffusion docking · FEP refinement

outputsSDF · ranked hit lists

Express compute intensity low: Single-cell and spatial transcriptomics, from count matrix to annotated atlas. Clustering, trajectory inference and cell-type labelling with reference atlases included.

scRNA-seq, snATAC & spatial data
Automatic cell-type annotation
Interactive embeddings you can share

built onGraph clustering · RNA velocity

outputsh5ad · atlases · UMAPs

the ritual

From Raw Data to Insight

Ingest

Connect S3, GCS or upload FASTA, FASTQ, VCF.

Compose

Chain modules into a versioned pipeline.

Run

GPUs spin up on demand, then spin down.

Publish

Export figures, notebooks and provenance.

Prefer code? The Python SDK speaks the same language as the workbench.

import mistfold as mf

project = mf.Project("kinase-screen")
target  = project.fold("P00533.fasta")            # predict EGFR structure
hits    = project.dock(target, library="enamine-1M", top_k=500)
hits.report().export("results/")               # figures + full provenance

the covenant

Plans

academic

Free

All four modules
200 GPU-hours / month
Public & private projects
Community support

Apply

lab

$490 / month

Everything in Academic
2,000 GPU-hours / month
10 seats, shared workspaces
Priority queue

Start trial

enterprise

Custom

Deploy in your own cloud
SSO, audit logs, GxP options
Unlimited seats
Dedicated scientist support

Talk to us

“We replaced three cluster scripts and a spreadsheet of parameters with one Mistfold pipeline. Our reviewers stopped asking how we got the numbers.” — Example testimonial · Structural biology lab