ColabFold

Pick ColabFold when you need a fast no-MSA fold. 1 to 2 min per run, no MMseqs2 round-trip. Pair with AF2 standalone (D2) when you want full MSA and templates, or with ESMFold (D4) for single-sequence monomers on an even smaller GPU.

ColabFold (Mirdita et al., Nature Methods 2022) running AlphaFold2 weights without MMseqs2 MSA fetch. Faster than full AF2 at the cost of MSA-derived accuracy. Useful when you need a structure quickly and the target has a tractable fold.

When it fits:

• You need a structure in 1 to 2 minutes and can tolerate slightly lower accuracy than full-MSA AF2.
• You're folding many sequences sequentially and need throughput.
• Your target is a well-folded monomer or small multimer with no exotic chemistry.

You will need:

• Single-letter FASTA sequence(s).
• Targets with deep evolutionary signal. Multi-domain or low-information sequences underperform without MSA.

Each run uses a preset that sets the scale and scope:

Standalone with your FASTA

Paste a FASTA (monomer or multimer up to 600 aa total) and get pLDDT, PAE, and pTM/ipTM. ~1 to 2 min on A100-40GB. No MSA, no templates. Pair with D2 AF2 if you need the full MSA-backed fold.

Batch for many fold targets

Fold many independent targets in one job (up to 200 records). Each record can be a monomer or a multimer (use ``:`` inside a record to break chains). Per-design results stream into the job page as folds complete. Fast no-MSA tier, ~1 to 2 min per fold.

Predicted PDB with per-residue pLDDT and PAE. Download as PDB or PAE matrix for filtering.

Where a tool reports them, the scores mean:

ipTM

Predicted confidence in the binder to target interface. Higher is better. Aim above roughly 0.7 on a tractable target.

pLDDT

Per-residue confidence in the predicted fold. Higher means the model is more sure of that part of the structure.

i_pAE and pAE

Predicted alignment error, at the interface (i_pAE) or across the whole structure (pAE). Lower is better.

Mirdita et al., Nature Methods 2022