SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.00002 · NeurIPS · 2024

Kermut: Composite kernel regression for protein variant effects

Peter Groth, Mads Kerrn, Lars Olsen, Jesper Salomon, Novonesis Wouter Boomsma

arXiv · PDF · Open in the Atlas

Code that ran

10 of 11 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; 1 honoured the contract we drafted, 2 violated it and 0 have no contract verdict; 7 failures of our instrument; 1 no recorded run), measured 2026-11-22; 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/2407.00002.json · badge: /api/ran/2407.00002.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
OATML-Markslab/ProteinGym canonical (classifier label)
classifier basis: census
10 ran · 1 no recorded run
FunctionStatusWhere it lives
RITA_gelu Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
OATML-Markslab/ProteinGym/proteingym/baselines/rita/rita_modeling.py
code served (permissive licence) · get_code("81f73e11236556b7")
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; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 torch.nn as nn

torch.manual_seed(0)

# The function RITA_gelu takes hidden_states as input and applies a GELU approximation
# hidden_states is used in mathematical operations, so it needs to be a tensor
hidden_states = torch.randn(2, 4)

ARGS = {
    "hidden_states": hidden_states
}
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 — no output shape is stated: observe_signature.py could not observe it (import:OSError)
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
calc_ndcg Ran
ran, honoured the contract we drafted
OATML-Markslab/ProteinGym/proteingym/performance_DMS_benchmarks.py
code served (permissive licence) · get_code("5c641396c1795538")
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 int, ndim 0, dtype family int, 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 numpy as np
import pandas as pd

# Set random seed for reproducibility
np.random.seed(0)

# Create sample data for y_true and y_score
# y_true: true scores (higher is better)
y_true = np.array([0.8, 0.3, 0.9, 0.1, 0.7, 0.5])

# y_score: predicted scores (higher is better)
y_score = np.array([0.75, 0.4, 0.85, 0.2, 0.65, 0.45])

ARGS = {
    'y_true': y_true,
    'y_score': y_score,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, 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
calc_toprecall Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
OATML-Markslab/ProteinGym/proteingym/performance_DMS_benchmarks.py
code served (permissive licence) · get_code("305bc10e974debbb")
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 float64, ndim 0, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 numpy as np

# Create sample score arrays
# The function expects numpy arrays and uses np.percentile and logical operations on them
true_scores = np.array([0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0])
model_scores = np.array([0.15, 0.25, 0.35, 0.45, 0.55, 0.65, 0.75, 0.85, 0.95, 0.99])

ARGS = {
    "true_scores": true_scores,
    "model_scores": model_scores,
    "top_true": 10,
    "top_model": 10
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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_bootstrap_standard_error Ran
ran; our contract draft was wrong, not the code
OATML-Markslab/ProteinGym/proteingym/performance_clinical_benchmarks.py
code served (permissive licence) · get_code("432d753249e4ae51")
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 float64, ndim 0, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 numpy as np
import pandas as pd

np.random.seed(0)

# Create a small DataFrame with AUC values per gene
# The function expects auc_per_gene to have a .values attribute
auc_data = {
    'gene_1': [0.75],
    'gene_2': [0.82],
    'gene_3': [0.68],
    'gene_4': [0.91],
    'gene_5': [0.79]
}
auc_per_gene = pd.DataFrame(auc_data).T
auc_per_gene.columns = ['auc']
auc_per_gene = auc_per_gene['auc']

ARGS = {
    'auc_per_gene': auc_per_gene,
    'n_bootstrap': 1000
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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_per_gene_auc Ran
ran; our contract draft was wrong, not the code
OATML-Markslab/ProteinGym/proteingym/performance_clinical_benchmarks.py
code served (permissive licence) · get_code("a0cb2321ae97ac2a")
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 os
import tempfile
import numpy as np
import pandas as pd

# Create a temporary directory for test score files
temp_dir = tempfile.mkdtemp()

# Create sample score CSV files
gene_ids = ['GENE1', 'GENE2', 'GENE3']
model_names = ['model_a', 'model_b']

for gene_id in gene_ids:
    # Create a dataframe with labels and model scores
    np.random.seed(0)
    n_samples = 20
    
    # Generate binary labels (Pathogenic/Benign)
    labels = np.random.choice(['Pathogenic', 'Benign'], size=n_samples)
    
    # Generate scores for each model
    data = {
        'DMS_bin_score': labels,
        'model_a': np.random.randn(n_samples),
        'model_b': np.random.randn(n_samples),
    }
    
    df = pd.DataFrame(data)
    score_file = os.path.join(temp_dir, f'{gene_id}.csv')
    df.to_csv(score_file, index=False)

# Define model configuration
model_config = {
    'model_a': {
        'directionality': 1,
        'model_type': 'neural_network'
    },
    'model_b': {
        'directionality': -1,
        'model_type': 'linear'
    }
}

# Define clean names mapping
clean_names = {
    'model_a': 'Model A',
    'model_b': 'Model B'
}

# Construct ARGS dictionary
ARGS = {
    'gene_ids': gene_ids,
    'model_names': model_names,
    'model_config': model_config,
    'scores_folder': temp_dir,
    'clean_names': clean_names
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, sklearn — 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
compute_pooled_auc Ran
ran; our contract draft was wrong, not the code
OATML-Markslab/ProteinGym/proteingym/performance_clinical_benchmarks.py
code served (permissive licence) · get_code("6445fcabd9a1ee84")
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 DataFrame. 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 os
import numpy as np
import pandas as pd
import tempfile

# Create a temporary directory for scores
temp_dir = tempfile.mkdtemp()

# Create sample score CSV files
gene_ids = ['GENE1', 'GENE2']
for gene_id in gene_ids:
    data = {
        'DMS_score_bin': ['Pathogenic', 'Benign', 'Pathogenic', 'Benign'],
        'model_A': [0.9, 0.1, 0.85, 0.2],
        'model_B': [0.8, 0.3, 0.75, 0.15],
    }
    df = pd.DataFrame(data)
    score_file = os.path.join(temp_dir, f'{gene_id}.csv')
    df.to_csv(score_file, index=False)

# Define model configuration
model_config = {
    'model_A': {
        'model_type': 'CNN',
        'directionality': 1
    },
    'model_B': {
        'model_type': 'RNN',
        'directionality': -1
    }
}

# Define clean names mapping
clean_names = {
    'model_A': 'Model A (CNN)',
    'model_B': 'Model B (RNN)'
}

ARGS = {
    'gene_ids': gene_ids,
    'model_names': ['model_A', 'model_B'],
    'model_config': model_config,
    'scores_folder': temp_dir,
    'clean_names': clean_names
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 and dtype family float describe that coercion, not the output, and are not 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
concat_csvs Ran
ran; our contract draft was wrong, not the code
OATML-Markslab/ProteinGym/proteingym/utils/download.py
code served (permissive licence) · get_code("763b7578ff85734d")
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 DataFrame. 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 os
import pandas as pd
import tempfile

# Create temporary CSV files to simulate pg_resources structure
temp_dir = tempfile.mkdtemp()

# Create sample CSV files
csv_files = []
for i in range(3):
    csv_path = os.path.join(temp_dir, f'clinical_sub_{i}.csv')
    df = pd.DataFrame({
        'protein_id': [f'prot_{j}' for j in range(2)],
        'substitution': [f'A{j}B' for j in range(2)],
        'score': [0.5 + j * 0.1 for j in range(2)]
    })
    df.index.name = 'index'
    df.to_csv(csv_path)
    csv_files.append(csv_path)

# Create pg_resources dictionary structure
pg_resources = {
    'clinical_ProteinGym_substitutions': csv_files
}

ARGS = {
    'pg_resources': pg_resources,
    'key': 'clinical_ProteinGym_substitutions',
    'max_files': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type DataFrame — a DataFrame output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataFrame output the recorded ndim 2 and dtype family float describe that coercion, not the output, and are not 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
minmax Ran
ran, violated the contract we drafted
OATML-Markslab/ProteinGym/proteingym/performance_DMS_benchmarks.py
code served (permissive licence) · get_code("1b86eba6fab689e0")
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 ndarray, ndim 1, dtype family float, shape (5,) 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 numpy as np

# The minmax function takes a single parameter 'x' which should be a numpy array
# It computes (x - min) / (max - min) for min-max normalization

# Create a small numpy array with distinct values to test the function
x = np.array([1.0, 2.0, 3.0, 4.0, 5.0])

ARGS = {
    "x": x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (5,) — shape (5,) 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
rotate_half Ran
ran; our contract draft was wrong, not the code
OATML-Markslab/ProteinGym/proteingym/baselines/rita/rita_modeling.py
code served (permissive licence) · get_code("2502f906145be7f6")
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 (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)

def rotate_half(x):
    x1, x2 = (x[..., :x.shape[-1] // 2], x[..., x.shape[-1] // 2:])
    return torch.cat((-x2, x1), dim=x1.ndim - 1)

# The function takes a single parameter 'x' which is a tensor
# It splits the last dimension in half, and concatenates (-x2, x1)
# We need a tensor with even last dimension

x = torch.randn(2, 4)

ARGS = {
    "x": x
}
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 (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
standardization Ran
ran, violated the contract we drafted
OATML-Markslab/ProteinGym/proteingym/merge.py
code served (permissive licence) · get_code("21fcd446fc722326")
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 ndarray, ndim 1, dtype family float, shape (5,) 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 numpy as np
import pandas as pd

# Create a small numpy array for testing
x_numpy = np.array([1.0, 2.0, 3.0, 4.0, 5.0])

ARGS = {
    "x": x_numpy
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (5,) — shape (5,) 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_bootstrap_standard_error_functional_categories Not yet run
unverified (harvested, no recorded run)
OATML-Markslab/ProteinGym/proteingym/performance_DMS_supervised_benchmarks.py
code served (permissive licence) · get_code("6f7c1afc63aecd20")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 1 later) and 9 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
rotate_half 9 other papers
7 dated earlier, 1 later, 1 same year, order not determinable
2305.11072 (2023-05)
Self-supervised Fine-tuning for Improved Content Representations by Speaker-invariant Clustering
2310.13802 (2023-10)
Improving Molecular Properties Prediction Through Latent Space Fusion
2311.04219 (2023-11)
OtterHD: A High-Resolution Multi-modality Model
and 6 more not named here
standardization 2 other papers
2 dated earlier
1702.05373 (2017-02)
EMNIST: an extension of MNIST to handwritten letters
2205.13760 (2022-05)
Tranception: protein fitness prediction with autoregressive transformers and inference-time retrieval

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

Reliable prediction of protein variant effects is crucial for both protein optimization and for advancing biological understanding. For practical use in protein engineering, it is important that we can also provide reliable uncertainty estimates for our predictions, and while prediction accuracy has seen much progress in recent years, uncertainty metrics are rarely reported. We here provide a Gaussian process regression model, Kermut, with a novel composite kernel for modeling mutation similarity, which obtains state-of-the-art performance for supervised protein variant effect prediction while also offering estimates of uncertainty through its posterior. An analysis of the quality of the uncertainty estimates demonstrates that our model provides meaningful levels of overall calibration, but that instance-specific uncertainty calibration remains more challenging.

For agents

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

get_harvested_code_for_paper("2407.00002")
get_code_for_paper("2407.00002")
have("2407.00002")

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

curl https://syntology.ai/api/ran/2407.00002.json

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

[![Syntology run record](https://syntology.ai/api/ran/2407.00002.svg)](https://syntology.ai/paper/2407.00002)

Connect an agent — have() is free.