SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.18813 · ICLR · 2024

Protein Multimer Structure Prediction via Prompt Learning

Hong Cheng, Jia Li, Yu Li, Ziqi Gao, Xiangguo Sun, Zijing Liu

arXiv · PDF · Open in the Atlas

Code that ran

10 of 16 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 6 failures of our instrument; 6 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2402.18813.json · badge: /api/ran/2402.18813.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
zqgao22/PromptMSP canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
zqgao22/promptmsp canonical (classifier label)
classifier basis: classifier_2025
0 ran · 1 no recorded run
sheldonresearch/ProG community
Papers with Code archive
4 ran · 3 no recorded run
FunctionStatusWhere it lives
BgGraphToSupernodePropagator Ran
ran on a synthesized input
sheldonresearch/ProG/prompt_graph/prompt/ProdigyPrompt.py
code served (permissive licence) · get_code("e2d2c7e3ea103917")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch import nn

def _scatter_mean(src, index, dim_size):
    """Pure-PyTorch replacement for ``torch_scatter.scatter_mean``."""
    out = torch.zeros(dim_size, src.size(1), device=src.device, dtype=src.dtype)
    count = torch.zeros(dim_size, 1, device=src.device, dtype=src.dtype)
    out.scatter_add_(0, index.unsqueeze(1).expand_as(src), src)
    count.scatter_add_(0, index.unsqueeze(1), torch.ones_like(index.unsqueeze(1), dtype=src.dtype))
    count = count.clamp(min=1)
    return out / count

def _scatter_sum(src, index, dim_size):
    """Pure-PyTorch replacement for ``torch_scatter.scatter_sum``."""
    out = torch.zeros(dim_size, src.size(1), device=src.device, dtype=src.dtype)
    out.scatter_add_(0, index.unsqueeze(1).expand_as(src), src)
    return out

class BgGraphToSupernodePropagator(nn.Module):
    """Up layer: aggregate background-graph nodes into supernode embeddings."""

    def __init__(self, aggr='mean'):
        super().__init__()
        self.aggr = aggr

    def forward(self, all_node_emb, supernode_edge_index, supernode_idx, graph_batch=None):
        if self.aggr == 'mean':
            aggr = _scatter_mean(all_node_emb[supernode_edge_index[0]], supernode_edge_index[1], dim_size=all_node_emb.size(0))
        else:
            aggr = _scatter_sum(all_node_emb[supernode_edge_index[0]], supernode_edge_index[1], dim_size=all_node_emb.size(0))
        return aggr[supernode_idx]


INIT_ARGS = {
    'aggr': 'mean'
}

ARGS = {
    'all_node_emb': torch.randn(6, 4),
    'supernode_edge_index': torch.tensor([[0, 1, 2, 3], [0, 0, 1, 1]], dtype=torch.long),
    'supernode_idx': torch.tensor([0, 1], dtype=torch.long),
    'graph_batch': None
}

model = BgGraphToSupernodePropagator(**INIT_ARGS)
output = model(**ARGS)
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
G_fn Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zqgao22/PromptMSP/dimer/cal_db5_rmsd_prody.py
code served (permissive licence) · get_code("3af636c75c65ce49")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 1, dtype family float, shape (2,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Analyze the function:
# - protein_coords: viewed as (1, -1, 3), so shape should be (N, 3) where N is number of atoms
# - x: viewed as (-1, 1, 3), so shape should be (M, 3) where M is batch size
# - sigma: a scalar float value
# The function computes distances between protein coordinates and batch points

# Create small tensors with distinct sizes
protein_coords = torch.randn(4, 3)  # 4 atoms, 3 coordinates each
x = torch.randn(2, 3)  # batch of 2 points, 3 coordinates each
sigma = 1.0  # scalar

ARGS = {
    "protein_coords": protein_coords,
    "x": x,
    "sigma": sigma,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float, shape (2,) — shape (2,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
SupernodeToBgGraphPropagator Ran
ran on a synthesized input
sheldonresearch/ProG/prompt_graph/prompt/ProdigyPrompt.py
code served (permissive licence) · get_code("931260bb5a363356")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned Tensor, ndim 2, dtype family float, shape (8, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch import nn

class SupernodeToBgGraphPropagator(nn.Module):
    """Down layer: project supernode embeddings back to the background graph."""

    def __init__(self, emb_dim):
        super().__init__()
        self.proj_sn_attr = nn.Linear(emb_dim, emb_dim)
        self.proj_sn_attr_2 = nn.Linear(emb_dim, emb_dim)

    def forward(self, x, new_supernode_x, supernode_edge_index, supernode_idx, graph_batch=None):
        x[supernode_idx] = x[supernode_idx] + self.proj_sn_attr(new_supernode_x)
        x[supernode_edge_index[0]] = x[supernode_edge_index[0]] + self.proj_sn_attr_2(x[supernode_edge_index[1]])
        return x


INIT_ARGS = {
    'emb_dim': 4,
}

ARGS = {
    'x': torch.randn(8, 4),
    'new_supernode_x': torch.randn(2, 4),
    'supernode_edge_index': torch.tensor([[0, 1, 2, 3], [4, 5, 6, 7]]),
    'supernode_idx': torch.tensor([0, 1]),
    'graph_batch': None,
}

if __name__ == '__main__':
    model = SupernodeToBgGraphPropagator(**INIT_ARGS)
    output = model(**ARGS)
    print(output.shape)
    print(output)
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (8, 4) — shape (8, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
_scatter_mean Ran
ran; our contract draft was wrong, not the code
sheldonresearch/ProG/prompt_graph/prompt/ProdigyPrompt.py
code served (permissive licence) · get_code("3b0b538e1b9053c2")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Construct arguments for _scatter_mean
# src: source tensor of shape (batch_size, feature_dim)
# index: index tensor of shape (batch_size,) with values in range [0, dim_size)
# dim_size: integer specifying output dimension size

src = torch.randn(2, 4)  # batch_size=2, feature_dim=4
index = torch.tensor([0, 1])  # indices for scattering, values in [0, 2)
dim_size = 2  # output will have shape (2, 4)

ARGS = {
    "src": src,
    "index": index,
    "dim_size": dim_size,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
_scatter_sum Ran
ran; our contract draft was wrong, not the code
sheldonresearch/ProG/prompt_graph/prompt/ProdigyPrompt.py
code served (permissive licence) · get_code("a5c95699b2163866")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Analyze the function:
# - src: a tensor that gets indexed and summed, used with .size(1) and .unsqueeze(1)
# - index: a tensor used to index src, gets .unsqueeze(1) and .expand_as(src)
# - dim_size: an integer used to create output tensor size

# src should be 2D: (batch_size, feature_dim)
# index should be 1D: (batch_size,) with values in range [0, dim_size)
# dim_size should be an integer

src = torch.randn(2, 4)  # 2 samples, 4 features
index = torch.tensor([0, 1])  # 2 indices, values in [0, 2)
dim_size = 2  # output will be (2, 4)

ARGS = {
    "src": src,
    "index": index,
    "dim_size": dim_size,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_body_intersection_loss Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zqgao22/PromptMSP/dimer/cal_db5_rmsd_prody.py
code served (permissive licence) · get_code("3b41ee1a4cc7e9b1")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create small tensors for ligand coordinates (deformed)
# Shape should be (N, 3) where N is number of ligand atoms
model_ligand_coors_deform = torch.randn(2, 3, dtype=torch.float32)

# Create small tensors for bound receptor representation nodes locations
# Shape should be (M, 3) where M is number of receptor nodes
bound_receptor_repres_nodes_loc_array = torch.randn(4, 3, dtype=torch.float32)

# Hyperparameters
sigma = 25.0
surface_ct = 10.0

ARGS = {
    "model_ligand_coors_deform": model_ligand_coors_deform,
    "bound_receptor_repres_nodes_loc_array": bound_receptor_repres_nodes_loc_array,
    "sigma": sigma,
    "surface_ct": surface_ct,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
find_rigid_alignment Ran
ran; our contract draft was wrong, not the code
zqgao22/PromptMSP/utils.py
code served (permissive licence) · get_code("0c03f954094c87f1")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create two point clouds for alignment
# A is the source point cloud (N=3 points, D=2 dimensions)
A = torch.tensor([[1., 1.], [2., 2.], [1.5, 3.]], dtype=torch.float)

# Create B by applying a rotation and translation to A
angle = np.pi / 3  # 60 degrees
R0 = torch.tensor([[np.cos(angle), -np.sin(angle)], [np.sin(angle), np.cos(angle)]], dtype=torch.float)
B = (R0.mm(A.T)).T
t0 = torch.tensor([3., 3.])
B = B + t0

ARGS = {
    "A": A,
    "B": B,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
generate_single_graph Ran
ran, honoured the contract we drafted
zqgao22/PromptMSP/produce_prompting_dgls.py
code served (permissive licence) · get_code("5e59a48de77c1fa7")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family int, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import random
import numpy as np

torch.manual_seed(0)
random.seed(0)
np.random.seed(0)

ARGS = {
    "chain_num": 10,
    "num": 5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family int, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
generate_single_graph Ran
ran, honoured the contract we drafted
zqgao22/PromptMSP/produce_training_dgls.py
code served (permissive licence) · get_code("58b7d52b321edd39")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family int, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import random
import numpy as np

torch.manual_seed(0)
random.seed(0)
np.random.seed(0)

ARGS = {
    "chain_num": 5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family int, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_rot_mat Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
zqgao22/PromptMSP/dimer/cal_db5_rmsd_prody.py
code served (permissive licence) · get_code("64ef8c256996fa77")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 2, dtype family float, shape (3, 3) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function expects euler_angles to be indexable with [0], [1], [2]
# It should contain three scalar values (roll, yaw, pitch)
# These are used with torch.cos and torch.sin, so they should be tensors or floats
euler_angles = torch.tensor([0.1, 0.2, 0.3])

ARGS = {
    "euler_angles": euler_angles
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (3, 3) — shape (3, 3) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
MetaGNN Not yet run
unverified (harvested, no recorded run)
sheldonresearch/ProG/prompt_graph/prompt/ProdigyPrompt.py
code served (permissive licence) · get_code("d92242abe79b15f5")
How it ran

no invocation recorded.

MetaGNNLayer Not yet run
unverified (harvested, no recorded run)
sheldonresearch/ProG/prompt_graph/prompt/ProdigyPrompt.py
code served (permissive licence) · get_code("7ca6d2564bc6ffdd")
How it ran

no invocation recorded.

ProdigyPrompt Not yet run
unverified (harvested, no recorded run)
sheldonresearch/ProG/prompt_graph/prompt/ProdigyPrompt.py
code served (permissive licence) · get_code("80c5d226eec14203")
How it ran

no invocation recorded.

assemble_rmsd_for_inference Not yet run
unverified (harvested, no recorded run)
zqgao22/PromptMSP/utils.py
code served (permissive licence) · get_code("9f29a0bfa605a6ef")
How it ran

no invocation recorded.

assemble_rmsd_for_inference Not yet run
unverified (harvested, no recorded run)
zqgao22/PromptMSP/utils_target.py
code served (permissive licence) · get_code("23edd3f838f93aec")
How it ran

no invocation recorded.

normalization Not yet run
unverified (harvested, no recorded run)
zqgao22/promptmsp/run_pre_training.py
code served (permissive licence) · get_code("61e7d9ad236ec9c1")
How it ran

no invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 4 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 2 later) and 12 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
G_fn 2 other papers
1 dated earlier, 1 later
2111.07786 (2021-11)
Independent SE(3)-Equivariant Models for End-to-End Rigid Protein Docking
2403.07902 (2024-03)
DECOMPDIFF: Diffusion Models with Decomposed Priors for Structure-Based Drug Design
compute_body_intersection_loss 1 other paper
1 dated earlier
2111.07786 (2021-11)
Independent SE(3)-Equivariant Models for End-to-End Rigid Protein Docking
find_rigid_alignment 1 other paper
1 later
2403.01694 (2024-03)
Tac-Man: Tactile-Informed Prior-Free Manipulation of Articulated Objects
get_rot_mat 1 other paper
1 dated earlier
2111.07786 (2021-11)
Independent SE(3)-Equivariant Models for End-to-End Rigid Protein Docking

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Understanding the 3D structures of protein multimers is crucial, as they play a vital role in regulating various cellular processes. It has been empirically confirmed that the multimer structure prediction (MSP) can be well handled in a step-wise assembly fashion using provided dimer structures and predicted proteinprotein interactions (PPIs). However, due to the biological gap in the formation of dimers and larger multimers, directly applying PPI prediction techniques can often cause a poor generalization to the MSP task. To address this challenge, we aim to extend the PPI knowledge to multimers of different scales (i.e., chain numbers). Specifically, we propose PROMPTMSP, a pre-training and Prompt tuning framework for Multimer Structure Prediction. First, we tailor the source and target tasks for effective PPI knowledge learning and efficient inference, respectively. We design PPI-inspired prompt learning to narrow the gaps of two task formats and generalize the PPI knowledge to multimers of different scales. We provide a meta-learning strategy to learn a reliable initialization of the prompt model, enabling our prompting framework to effectively adapt to limited data for large-scale multimers. Empirically, we achieve both significant accuracy (RMSD and TM-Score) and efficiency improvements compared to advanced MSP models. The code, data and checkpoints are released at https://github.com/zqgao22/PromptMSP.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2402.18813")
get_code_for_paper("2402.18813")
have("2402.18813")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2402.18813.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2402.18813.svg)](https://syntology.ai/paper/2402.18813)

Connect an agent — have() is free.