ESMFold

Meta AI’s protein structure prediction from single sequence (no MSA needed). Faster than AlphaFold, useful for quick validation and disorder prediction.

What It Does

• Predicts 3D protein structure from amino acid sequence alone
• No multiple sequence alignment needed (faster than AF2)
• Per-residue pLDDT confidence scores
• Good at identifying disordered regions

How to Use

API

curl -X POST https://api.esmatlas.com/foldSequence/v1/pdb/ \
  -H "Content-Type: text/plain" \
  -d "MVLSPA..." > prediction.pdb

Limitation

Public API has 400 aa limit. STRC is 1775 aa — must split into segments.

Python

import requests
seq = "YOUR_SEQUENCE_HERE"  # max 400 aa
r = requests.post("https://api.esmatlas.com/foldSequence/v1/pdb/",
    data=seq, headers={"Content-Type": "text/plain"})
with open("prediction.pdb", "w") as f:
    f.write(r.text)

Verified Status

VERIFIED — used for STRC disorder validation. Submitted 3 segments due to 400 aa limit. Cross-validated with AlphaFold pLDDT and IUPred3. E1659 falls in segment 3 (ordered region) — ESMFold confirms it’s well-structured, consistent with AF pLDDT 95.69.

STRC Research Usage

• STRC ESMFold Disorder Validation — validated AlphaFold disorder predictions
• Cross-validated N-terminal disorder (confirmed disordered region safe to truncate for mini-STRC)

Connections

• AlphaFold Database [see-also] — higher accuracy but slower
• IUPred3 [see-also] — complementary disorder prediction
• STRC ESMFold Disorder Validation [used-in] 55|- STRC pLDDT Profile and Cut Point Analysis [see-also]

STRC E1659A Results

Previously tested for STRC full-length structure prediction. Tested: 2026-04-08