IgGM

IgGM is a no-install online antibody and nanobody engineering tool you can run through tools.ranomics.com on a dedicated GPU. One model for the antibody work that usually takes four separate ones — loop redesign, humanisation, affinity maturation and predicting the docked complex all run from the same upload.

Example run — Hen egg-white lysozyme, chain A of PDB 1HEW, 129 residues — with a nine-residue epitope named on its surface: 63 to 68 and 73 to 75. See what came back →

Design inputs

Mode

Antibody-antigen complex prediction Fold the antibody-antigen complex from the full heavy (and light) chain sequences against your antigen. Provide the epitope residues to guide docking. Fastest mode. CDR design (H3 or all CDRs) Mask CDR positions with X in the antibody FASTA and IgGM redesigns them against your antigen, aimed at the epitope residues you specify. Framework redesign / humanization Mask framework positions with X; IgGM redesigns them. Use for humanization or framework engineering.

Antibody FASTA

Paste the heavy chain as >H and, for a conventional antibody, the light chain as >L. Omit >L for a nanobody / VHH. Mark positions to design with X. Do not include an antigen record; it is read from the uploaded PDB.

Antigen PDB

Upload the target as .pdb / .cif. The antigen sequence is read from the chain you select, so the structure is the single source of truth.

Antigen chain

Epitope residues (required)

Required: the antigen residue numbers IgGM should target (space or comma separated, PDB numbering). IgGM designs against a specific epitope and cannot infer one from an antigen-only structure. Get them from Epitope Scout or your own structural analysis. Positions are handled correctly regardless of the PDB's residue numbering.

Advanced

Samples

Max antigen size

Samples: designs to generate (up to 100 total). For affinity maturation, samples are counted per masked position, so samples x masked positions must stay at or under 100 (needs at least 2 samples). Max antigen size: IgGM's memory guard; lower it for very large antigens.

Group

Group related runs on your jobs page. Leave blank to file under Uncategorized.

You pay for the GPU time your run uses. Cancelling does not refund GPU time already used.

About IgGM

What it is

Takes an antibody or nanobody and an antigen and does whatever you need to the pair: redesign the binding loops (the CDRs), rebuild or humanise the framework around them, or simply predict how the two dock. All of it is aimed at the epitope you name, and all of it comes out of one model rather than a chain of them. IgGM, Wang et al., ICLR 2025.

When to use

You already have an antibody or nanobody and want its binding loops, or its framework, rebuilt against a specific antigen and epitope.
You want to humanise a framework, or rebuild it against a new antigen.
You want to see how an antibody sits on its antigen before committing to a wet-lab project.

Prerequisites

Antigen structure as PDB or mmCIF (the antigen sequence is read from this file — you do not type it).
Antibody heavy chain sequence (>H); light chain (>L) optional (omit it for a nanobody / VHH). Mark positions to design with X.
An epitope: the antigen residues the antibody should aim at. Required for every mode — tools/iggm/__init__.py rejects a run without one.

Input glossary

Antibody FASTA. Paste the heavy chain as >H and, for a conventional antibody, the light chain as >L. Mark residues to design with X. Omit >L for a nanobody / VHH. Do not include an antigen record — it comes from the uploaded PDB.
Antigen PDB. Upload the target as .pdb / .cif. The antigen sequence is extracted from the chain you select, so the structure is the single source of truth.
Antigen chain. The chain ID in the uploaded PDB that IgGM should treat as the antigen (e.g. A).
Epitope. Required. The antigen residues IgGM should aim the design at; it cannot infer them from an antigen-only structure. Positions are handled correctly regardless of the PDB's residue numbering.
Mode. Complex prediction folds the complex; CDR design / framework redesign redesign masked (X) positions.

Per design: the predicted antibody-antigen complex PDB, the designed sequence, and an epitope-contact count (how many of your chosen epitope residues the designed antibody engages). Sequence statistics and amino-acid distribution plots are attached as artifacts.

References

A guided first run

Before designing anything, check that IgGM can place your existing antibody on your antigen. A single sample in prediction mode is already the smallest run IgGM offers, so these are the form’s own defaults — a guided first run, not a reduced-scope trial.

What you need

Your antigen structure file, and your antibody heavy chain sequence. The light chain is optional — omit it for a nanobody.

Hotspot residues are the numbered residues on the patch of your target you want the binder to sit on, written in your PDB’s own numbering — this tool will not start without at least one. If you do not know yours yet, score your target’s surface with Epitope Scout first — it is free, and its results hand the target and the residues back into this form.

What it costs

You pay for the GPU time your run uses.

Then what

If the predicted complex looks right, switch the mode to CDR design and mark the positions you want redesigned with X.

Load these settings →

A past run we actually did

The target

Hen egg-white lysozyme, chain A of PDB 1HEW, 129 residues — with a nine-residue epitope named on its surface: 63 to 68 and 73 to 75.

IgGM is not a de novo binder tool, so a worked example needs a real antibody as well as a real antigen. It takes an antibody you already have and rewrites its CDR loops; there is nothing for it to do without one. Lysozyme is the textbook antigen for exactly this work — small, rigid and thoroughly characterised — and the nine residues are one contiguous surface patch, which is the shape of thing you would type if you had mapped an epitope and wanted the loops aimed at it.

What we put in the form

Mode — CDR design: Redesigns the loops and leaves the framework alone. The other modes rebuild the framework instead, or only predict the complex without designing anything.
Antibody FASTA — two records, H (114 aa) and L (107 aa): The heavy chain carries five X characters, and X is the mask: those are the positions IgGM is free to rewrite. A design mode given an input with no X in it has nothing to design.
Antigen PDB — the 1HEW file: Chain A is the lysozyme, numbered 1-129 as it downloads.
Antigen chain — A: Names the chain the epitope numbers below refer to.
Epitope residues (required) — 63 64 65 66 67 68 73 74 75: Space separated, in the file's own numbering: Trp63, Cys64, Asn65, Asp66, Gly67 and Arg68, then Arg73, Asn74 and Leu75. Worth knowing before you read the result — this is a request, not a constraint. Nothing forces a design to land here.
Samples — 40: Forty independent designs. Worth this many because the outcome per design is close to all-or-nothing, so the useful answer is the spread across a batch rather than any one design.

What came back

40 designs, each scored by how many of the nine requested epitope residues it actually contacts. Nineteen of the forty reach no part of the epitope. Fourteen reach exactly one residue. The rest go further — three touch two, one touches four, two touch five, and the best design touches six of nine.

IgGM design cdr_design

Antigen

chain A, 129 aa

Epitope

63, 64, 65, 66, 67, 68, 73, 74, 75

Each design's epitope-contact count is how many of your requested epitope residues the designed antibody engages in the predicted complex (heavy atom within 5 Å). A design that engages the epitope is docking where you asked.

Candidates (40) CDR design

Returned 40 / 40

#	Epitope contacts ?	3D	Structure
1 Top	6	—	—

2	5	—	—

3	5	—	—

4	4	—	—

5	2	—	—

6	2	—	—

7	2	—	—

8	1	—	—

9	1	—	—

10	1	—	—

11	1	—	—

12	1	—	—

13	1	—	—

14	1	—	—

15	1	—	—

16	1	—	—

17	1	—	—

18	1	—	—

19	1	—	—

20	1	—	—

21	1	—	—

22	0	—	—

23	0	—	—

24	0	—	—

25	0	—	—

26	0	—	—

27	0	—	—

28	0	—	—

29	0	—	—

30	0	—	—

31	0	—	—

32	0	—	—

33	0	—	—

34	0	—	—

35	0	—	—

36	0	—	—

37	0	—	—

38	0	—	—

39	0	—	—

40	0	—	—

Validating in the lab is optional Designs from your own run are yours to download and take anywhere. If you do want them made and measured, Ranomics runs yeast display, mammalian display, BLI and DMS on the same data loop.

Test binders in the lab AI Binder Sprint

Epitope contacts (top 12 designs)

Columns are the epitope residues you asked for, by the number in your PDB. IgGM counts positions along chain A from 1, so A63 is position 63 inside the run.

Design	A63 pos 63	A64 pos 64	A65 pos 65	A66 pos 66	A67 pos 67	A68 pos 68	A73 pos 73	A74 pos 74	A75 pos 75	Hits
010_design_18										6 / 9
018_design_25										5 / 9
019_design_26										5 / 9
021_design_28										4 / 9
011_design_19										2 / 9
020_design_27										2 / 9
031_design_37										2 / 9
001_design_1										1 / 9
002_design_10										1 / 9
003_design_11										1 / 9
004_design_12										1 / 9
005_design_13										1 / 9

Artifacts

design_0.fasta
design_1.fasta
design_10.fasta
design_11.fasta
design_12.fasta
design_13.fasta
design_14.fasta
design_15.fasta
design_16.fasta
design_17.fasta
design_18.fasta
design_19.fasta
design_2.fasta
design_20.fasta
design_21.fasta
design_22.fasta
design_23.fasta
design_24.fasta
design_25.fasta
design_26.fasta
design_27.fasta
design_28.fasta
design_29.fasta
design_3.fasta
design_30.fasta
design_31.fasta
design_32.fasta
design_33.fasta
design_34.fasta
design_35.fasta
design_36.fasta
design_37.fasta
design_38.fasta
design_39.fasta
design_4.fasta
design_5.fasta
design_6.fasta
design_7.fasta
design_8.fasta
design_9.fasta

How to read that

There is one score column here, and reading it as a leaderboard throws away most of what this run told you. Two other things matter more. First, the shape. Nearly half the designs miss the epitope completely. That is an ordinary IgGM result, and it is why forty samples is the right order of magnitude rather than four: a run of four could easily have come back four zeros and told you nothing about your antibody or your epitope. Second, the contact grid below the table, which is the part worth the most and the part with no number attached. Its columns are your nine requested positions, and three of them — 63, 64 and 66 — are empty for every single design. Not rarely hit: never hit. Meanwhile position 73 is contacted by seventeen of the forty, and is the only contact five of them make. So the run is not sampling your nine-residue patch evenly. It is repeatedly finding one reachable sub-patch — 65, 67, 68, 73, 74 and 75 — while three residues sit somewhere a CDR loop on this framework does not get to. That is a finding about your epitope definition rather than about any design, and no ranked column would ever have shown it to you.

What happened next

Took the empty columns as the actionable result, which is not where you would look if you had gone straight to the top row. The next run narrows the epitope request to the residues that proved reachable, so the sampling concentrates there instead of spending part of every design on positions that never come back. Folding the best design independently and taking it to expression is the step after that, not before it.

Recorded from that one past run. That is what it did, not a prediction for yours.

What good looks like

Each tool reports a subset of these depending on whether it does design, sequence recovery, or structure prediction.

ipTM: Predicted confidence in the contact between two chains, on a 0 to 1 scale. Higher is better: 0.65 to 0.75 depending on the tool. Individual tools set their own pass bar a little either side of that. On a multi-chain target the number may cover the target's own chain–chain interface as well as the binder's.
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.
ProteinMPNN recovery: Fraction of native residues recovered when ProteinMPNN redesigns a known sequence on its native backbone — the folded shape on its own, before a sequence is chosen for it. Higher is better: > 0.4 usable; > 0.6 excellent.

Related tools on Ranomics

If you are picking between IgGM and a sibling algorithm, these run on the same hub against the same target.

Run RFantibody online: You have a target structure and want nanobodies — single-domain antibodies you can carry straight into yeast display, mammalian display or a hybridoma workflow. Each candidate is refolded and scored against the target before you see it. For non-antibody mini-proteins use RFdiffusion or BindCraft; for sugar-coated targets use BoltzGen.
Run BoltzGen online: You have a target and have not settled on what shape the binder should be. One model here aims mini-proteins, nanobodies, antibodies or peptides at the same site, so you can compare formats instead of guessing. The only design tool here that handles sugars and modified residues on the target natively.
Run Boltz-2 online: You already have a binder sequence and the target it should hit, and you want to know whether they actually stick together before you order DNA. Returns the predicted complex and a 0-to-1 interface confidence score. Trained on antibody-antigen complexes, so it gives a genuine second opinion next to AlphaFold2.