Home Research Services
Online Services
STOPred
StoPred result

example

model_trainingModelCurrent release (cutoff 2026-01-01)scheduleGenerated2026-10-28 03:23arrow_backBack to StoPred
Predicted stoichiometry
A1B6
Ranking score 0.49
Next: A3B6 (0.08), A1B2 (0.08)
1AH1227-16BH1227-2
Copies12345-67+
Top PDB template: 9N8W assembly 1 · A×1 · B×6 · see Templates
Co-assembly prediction
Score 0.9991
Drag the red threshold to see the trade-off
0.787
0.10.50.750.90.990.999
Recall
72%
Precisioninfo_outlinePrecision depends on how common true complexes are among the pairs tested. The benchmark has 555 complexes among 1,426 pairs (39%); precision is lower when complexes are rarer among the pairs you test.
65%
False-positive rate
25%

Estimated on curated benchmark pairs.

Default 0.787 maximizes macro-F1 on validation data (release 2026-01-01).

Stoichiometry ranking

RankStoichiometryRanking scoreCopies per chain
1A1B6
0.4931
A H1227-1: 1; B H1227-2: 6
2A3B6
0.0814
A H1227-1: 3; B H1227-2: 6
3A1B2
0.0771
A H1227-1: 1; B H1227-2: 2
4A2B6
0.0753
A H1227-1: 2; B H1227-2: 6
5A6B6
0.0613
A H1227-1: 6; B H1227-2: 6
6A7B6
0.0411
A H1227-1: 7; B H1227-2: 6
7A8B6
0.0402
A H1227-1: 8; B H1227-2: 6
8A5B6
0.0387
A H1227-1: 5; B H1227-2: 6
9A1B5
0.0168
A H1227-1: 1; B H1227-2: 5
10A12B6
0.0134
A H1227-1: 12; B H1227-2: 6

Per-chain copy numbers

Probability of each copy number for every chain, from the chain-level head. "other" is a copy number outside the trained classes.

AH1227-1427 aa
1
0.627
6
0.107
2
0.091
3
0.072
4
0.014
7
0.014
BH1227-2877 aa
6
0.931
2
0.036
3
0.009
4
0.006
7
0.006
12
0.004

Global Stoichiometry Predictions

Overall stoichiometry classes based on copy number ratios. For example, both A2B1 and A1B2 configurations belong to the (2,1) class.

(6,1)
0.403
(2,1)
0.132
(6,3)
0.096
(6,2)
0.066
(5,1)
0.033
(6,6)
0.023
(12,6)
0.015
(24,24)
0.012

Templates

Sequence search (MMseqs) of the submitted chains against PDB biological assemblies (template database StoPred_template_db_20260505, PDB snapshot 2026-05-05). Query chains that a template deposits as one chain (merge) or as several chains (split) are matched as one unit. 16 templates found. Click a row for the structure and the alignments.

Limited to PDB entries released before 2026-01-01 (0 later templates left out).

#TemplateScoreIdentityE-valueQuery positionTemplate positionTemplate copy numbers
19N8W asm 1
2571A 100%
B 100%
5e-282
A
B
A×1 · B×6
29N8X asm 1
2571A 100%
B 100%
5e-282
A
B
A×1 · B×6
36V0C asm 1
784B 100%0
A
B
B×6
46V0D asm 1
784B 100%0
A
B
B×6
56V0E asm 1
784B 100%0
A
B
B×6
66V0F asm 1
784B 100%0
A
B
B×6
76V0G asm 1
784B 100%0
A
B
B×6
86V0H asm 1
784B 100%0
A
B
B×6
96V0I asm 1
784B 100%0
A
B
B×6
106V0J asm 1
784B 100%0
A
B
B×6
117SEE asm 1
568B 86%0
A
B
B×6
127SEF asm 1
568B 86%0
A
B
B×6
136KZ3 asm 1
244B 100%5e-295
A
B
B×6
146KZ4 asm 1
152B 100%2e-228
A
B
B×6
156VCI asm 1
58B 98%2e-132
A
B
B×1
165UVN asm 1
7B 31%9e-33
A
B
B×6

Input sequences

AH1227-1427 aa
1MALNTPQITPTKKITVRAIGEELPRGDYQRCPQCDMLFSLPEINSHQSAYCPRCQAKIRD
61GRDWSLTRLAAMAFTMLLLMPFAWGEPLLHIWLLGIRIDANVMQGIWQMTKQGDAITGSM
121VFFCVIGAPLILVTSIAYLWFGNRLGMNLRPVLLMLERLKEWVMLDIYLVGIGVASIKVQ
181DYAHIQAGVGLFSFVALVILTTVTLSHLNVEELWERFYPQRPATRRDEKLRVCLGCHFTG
241YPDQRGRCPRCHIPLRLRRRHSLQKCWAALLASIVLLLPANLLPISIIYLNGGRQEDTIL
301SGIMSLASSNIAVAGIVFIASILVPFTKVIVMFTLLLSIHFKCQQGLRTRILLLRMVTWI
361GRWSMLDLFVISLTMSLINRDQILAFTMGPAAFYFGAAVILTILAVEWLDSRLLWDAHES
421GNARFDD
BH1227-2877 aa
1MSQETPASTTEAQIKNKRRISPFWLLPFIALMIASWLIWDSYQDRGNTVTIDFMSADGIV
61PGRTPVRYQGVEVGTVQDISLSDDLRKIEVKVSIKSDMKDALREETQFWLVTPKASLAGV
121SGLDALVGGNYIGMMPGKGKEQDHFVALDTQPKYRLDNGDLMIHLQAPDLGSLNSGSLVY
181FRKIPVGKVYDYAINPNKQGVVIDVLIERRFTDLVKKGSRFWNVSGVDANVSISGAKVKL
241ESLAALVNGAIAFDSPEESKPAEAEDTFGLYEDLAHSQRGVIIKLELPSGAGLTADSTPL
301MYQGLEVGQLTKLDLNPGGKVTGEMTVDPSVVTLLRENTRIELRNPKLSLSDANLSALLT
361GKTFELVPGDGEPRKEFVVVPGEKALLHEPDVLTLTLTAPESYGIDAGQPLILHGVQVGQ
421VIDRKLTSKGVTFTVAIEPQHRELVKGDSKFVVNSRVDVKVGLDGVEFLGASASEWINGG
481IRILPGDKGEMKASYPLYANLEKALENSLSDLPTTTVSLSAETLPDVQAGSVVLYRKFEV
541GEVITVRPRANAFDIDLHIKPEYRNLLTSNSVFWAEGGAKVQLNGSGLTVQASPLSRALK
601GAISFDNLSGASASQRKGDKRILYASETAARAVGGQITLHAFDAGKLAVGMPIRYLGIDI
661GQIQTLDLITARNEVQAKAVLYPEYVQTFARGGTRFSVVTPQISAAGVEHLDTILQPYIN
721VEPGRGNPRRDFELQEATITDSRYLDGLSIIVEAPEAGSLGIGTPVLFRGLEVGTVTGMT
781LGTLSDRVMIAMRISKRYQHLVRNNSVFWLASGYSLDFGLTGGVVKTGTFNQFIRGGIAF
841ATPPGTPLAPKAQEGKHFLLQESEPKEWREWGTALPK
How to read this page
  • Stoichiometry assumes the submitted sequences form one complex and ranks copy numbers per chain (A2B1 = two copies of chain A, one of chain B). It is always reported.
  • Assembly-state check tests that assumption. For one sequence, a monomer-aware model estimates whether it is a monomer or a homo-oligomer. For 2–6 distinct sequences, a co-assembly model scores whether they occur in one assembly.
  • Scores are model estimates from sequence alone (ESM-C embeddings), not experimental evidence.
Liu, Q., Peng, C., Zheng, W., Zhang, C., & Freddolino, L. (2025). StoPred: Accurate Stoichiometry Prediction for Protein Complexes Using Protein Language Models and Graph Attention. bioRxiv. https://doi.org/10.1101/2025.10.20.683515

FreddolinoLabumich.edu | 1150 W. Medical Center Dr, Ann Arbor, MI 48109-0600