Protein–protein interaction — two complementary panels. Left: evidence-based candidate ranking from local data (co-expression, co-localisation, KEGG, Pfam). Right: structure-model channel — package inputs for AlphaFold3 Server or ColabFold (AF2-Multimer), then upload the result archive to read ipTM / pTM / PAE. No folding runs on this server.

1 · Evidence screening

TWCG × TWCG only · local data only
Works only within TWCG (co-expression, co-localisation, shared KEGG pathways, shared Pfam domains). For TWCG × Cs cross-species pairs there is no shared evidence — use the structure panel on the right.
T1 = TWCG_HapA1.016779 (LRR-RLP, plasma membrane)
Scoring candidates…

2 · Structure model (AF3 / ColabFold)

no compute here — package & parse
Either chain can be a TWCG gene, a Cs (Ciboria shiraiana) gene, or a pasted sequence — cross-species complexes (TWCG × Cs) are fully supported.
  1. Put AlphaFold3 Server first: upload the JSON at alphafoldserver.com (free, Google login), run, download the output zip.
  2. No cluster time? The ColabFold pack ships FASTA + run script + Slurm template for any GPU box (AF2-Multimer v3, 5 models, 3 recycles).
  3. Bring the result back to Parse result — ipTM / pTM / PAE and the complex structure are read here.
Folding inference is not run on this page: structures are computed by AlphaFold3 Server or by ColabFold on GPU/HPC, and this site only packages the input and parses the results.