OSCR

A prototype-augmented graph representation learning framework for identifying brain disorder-associated genes and facilitating drug repurposing.

Code ↔ Paper

4 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 4 matches
  1. [1] § Methods › Data source and processing ↔ Tutorial.ipynb, lines 1–27 · score 0.87 · Parietal Lobe, Occipital Lobe, Temporal Lobe, Frontal Lobe, multi omics, SNP interaction
  2. [2] § Methods › Data source and processing ↔ data_preprocess_cv.py, lines 174–207 · score 0.83 · Parietal Lobe, Occipital Lobe, Temporal Lobe, Frontal Lobe, preprocessing, Cerebellum
  3. [3] § Methods › Training the model ↔ main.py, lines 50–96 · score 0.69 · Focal loss, training mask, neighbors, epochs, unlabeled, batch
  4. [4] § Methods › Transformer-based gene prediction model for brain disorders ↔ model.py, lines 25–112 · score 0.59 · layer normalization, dropout, linear, max, transformer, graph

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

Jupyter notebook · 141 lines · 5.8 KB · no license · 1 match

  1. # %% [markdown]
  2. # # Tutorial
  3. # This tutorial demonstrates how to use MOGT functions with a demo dataset (SCZ as an example). Once you are familiar with SCZ’s workflow, please replace the demo data with your own data to begin your analysis.
  4. #
  5. # ## How to prepare input data
  6. #
  7. # We recommend getting started with SCZ using the provided demo dataset. When you want to apply SCZ to your own multi-omics dataset, please refer to the following tutorials to learn how to prepare input data.
  8. #
  9. # Overall, the input data consists of two parts: the graph, constructed from SNP-SNP interaction and the node feature including DE, EPI, and gene expression in five brain regions（Parietal Lobe, Frontal Lobe, Temporal Lobe, Cerebellum, and Occipital Lobe）in adolescents and adults.
  10. #
  11. # If you are unfamiliar with MOGT, you may start with our data used in the paper to save your time. For SCZ, the input data as well as the label information are uploaded [here](https://github.com/NBStarry/CGMega/tree/main/data). If you start with this data, you can skip the _step 1_ about _How to prepare input data_.
  12. #
  13. # > The labels should be collected yourself if you choose analyze your own data.
  15. # %% [markdown]
  16. # ### Import default params and set constants
  18. # %% [markdown]
  19. # ### load omics data
  20. # The input demo data as well as the label information are uploaded [here](https://github.com/JiafangLi/MOGT/tree/main/data), put it into DATA_DIR, and run:
  22. # %%
  23. node_feat, pos = get_node_feat(disease=disease)
  24. feat_cols = ["de","brain_cp","brain_gz","neuron_count","oligo_count","micro_count","adolescence_parietal.lobe","adolescence_frontal.lobe","adolescence_temporal.lobe","adolescence_cerebellum","adolescence_occipital.lobe","adulthood_parietal.lobe","adulthood_frontal.lobe","adulthood_temporal.lobe","adulthood_cerebellum","adulthood_occipital.lobe"]
  25. feat_df = pd.DataFrame(data=node_feat, columns=feat_cols)
  26. feat_df = pd.concat([GENE_LIST, feat_df], axis=1)
  27. feat_df[:10]
  29. # %% [markdown]
  30. # ### SNP-SNP interaction construction
  31. # Then, we read the SNP-SNP interaction data (SNP-SNP interaction from [Zenodo]()) and transform it into a graph through the following commands:
  33. # %%
  34. snp_mat = get_snp_mat(disease=disease)
  35. snp_mat[20:30, 20:30]
  37. # %% [markdown]
  38. # ### Load gene labels
  40. # %%
  41. node_lab, labeled_idx = get_label(disease=disease )
  42. labeled_lab = [node_lab[i][1] for i in labeled_idx]
  43. print(f"Positive samples: {sum(node_lab)[1]} Negative samples: {sum(node_lab)[0]} Unlabeled samples: {len(node_lab) - len(labeled_idx)}")
  44. print(node_lab)
  46. # %% [markdown]
  47. # ### Build PyG Data instance
  49. # %%
  50. from data_preprocess_cv import build_pyg_data
  51. edge_threshhold = configs["edge_threshhold"]
  52. random = configs["random"]
  53. data = build_pyg_data(GENE_LIST,node_feat, node_lab, snp_mat, pos,edge_threshhold,random)
  54. data
  56. # %% [markdown]
  57. # ### Do train-test split
  58. #
  59. # 25% for test set, 75% for a 10-fold train-valid spilt.
  61. # %%
  62. from data_preprocess_cv import split_data
  63. train_idx_list,valid_idx_list,test_idx = split_data(CV_FOLDS,labeled_idx,labeled_lab)
  65. # %% [markdown]
  66. # ### Create CancerDataset and Save
  68. # %%
  69. import pickle
  70. cv_dataset = create_cv_dataset(train_idx_list.copy(), valid_idx_list.copy(), test_idx.copy(), data=data)
  71. print(cv_dataset[0])
  72. dataset_dir = "data/" + configs["disease"] + "/"+ configs["disease"]+ "_dataset.pkl"
  73. print(f'Finished! Saving dataset to {dataset_dir} ......')
  74. with open(dataset_dir, 'wb') as f:
  75. pickle.dump(cv_dataset, f)
  77. # %% [markdown]
  78. # ### Load data and Train model <div id="load-data-and-train-model"></div>
  79. #
  80. # A 5-fold training.
  82. # %%
  83. from main import get_training_modules, train_model, predict, calculate_metrics, pred_to_df,test
  84. from data_preprocess_cv import scale_data
  86. configs['stable'] = True
  88. # set to 'cuda' to use GPU.
  89. configs['device'] = 'cuda'
  91. # Set log_name and logfile to suit your needs, and you will see training details in logfile.
  92. configs['log_name'] = configs["disease"]
  93. configs['logfile'] = os.path.join(configs["log_dir"], configs["log_name"] + ".txt")
  95. dataset = get_data(configs,disease)
  98. def calculate_confidence_interval(scores):
  99. mean = np.mean(scores)
  100. std = np.std(scores)
  101. n = len(scores)
  102. z = 1.96 # Z-value at 95% confidence level
  103. bound = (z * std / np.sqrt(n))
  104. return bound
  106. sum_auprc, sum_auc, sum_acc, sum_f1, sum_tp, train_result = [], [], [], [], [], []
  107. for i in range(CV_FOLDS):
  108. head_info = True if i == 0 else False
  109. configs['fold'] = i
  110. data = dataset
  111. data = scale_data(data,i)
  112. modules = get_training_modules(configs, data)
  113. auprc, auc, acc, f1, tp, new_ckpt,cutoff = train_model(modules, configs, configs['log_name'], i, head_info,)
  114. y_score, y_pred, y_true, y_index, genes = predict(modules['model'], modules['test_loader_list'], configs, new_ckpt)
  115. acc, cf_matrix, auprc, f1, auc, test_cutoff = calculate_metrics(y_true, y_pred, y_score)
  116. #save test result
  117. y = pd.DataFrame({'y_true': y_true, 'y_score': y_score,"y_pred":y_pred,"cutoff":cutoff})
  118. result = pd.concat([genes,y], axis=1)
  119. result.to_csv(configs["out_dir"]+"/"+disease +"/"+disease+"_"+"test"+"_"+str(i)+".csv",index_label=False)
  122. tp = cf_matrix[1, 1]
  123. sum_auprc.append(auprc)
  124. sum_auc.append(auc)
  125. sum_acc.append(acc)
  126. sum_f1.append(f1)
  127. sum_tp.append(tp)
  128. with open(configs['logfile'], 'a') as f:
  129. print("Test AUPRC:{:.4f}, AUROC:{:.4f}, ACC:{:.4f}, F1:{:.4f}, TP:{:.1f},cutoff:{:.4f}"
  130. .format(auprc, auc, acc, f1, tp,test_cutoff), file=f, flush=True)
  131. train_result = pred_to_df(i, train_result, y_index, y_true, y_score)
  132. auc_ci = calculate_confidence_interval(sum_auc)
  133. auprc_ci = calculate_confidence_interval(sum_auprc)
  134. avg_auc = sum(sum_auc) / len(sum_auc)
  135. avg_Auprc = sum(sum_auprc) / len(sum_auprc)
  136. avg_fi = sum(sum_f1) / len(sum_f1)
  137. with open(configs['logfile'], 'a') as f:
  138. print(f"{CV_FOLDS}-folds AUPRC:{avg_Auprc:.4f}+{auprc_ci}, AUROC:{avg_auc:.4f}+{auc_ci}, F1:{avg_fi:.4f}",
  139. file=f, flush=True)

Tutorial.ipynb at commit 25ff432, no license · at the source

Overview

Authors: Jiafang Li1,2, Yifei Li1,2, Siying Lin1,3, Jiahua Rao3, Huiying Zhao1,2
ORCID iDs: Jiafang Li, Huiying Zhao
  1. Department of Medical Research Center, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, Guangzhou, Guangdong, China
  2. Guangdong Provincial Key Laboratory of Malignant Tumor Epigenetics and Gene Regulation, Guangzhou, Guangdong, China
  3. School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, Guangdong, China
Institutions: Sun Yat-sen University (China); Sun Yat-sen Memorial Hospital (China)
Journal: PLoS computational biology, volume 22, issue 5, article e1014323
Dates: received 1 December 2025; accepted 11 May 2026; published online 29 May 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.1371/journal.pcbi.1014323 · PMID 42213730 · PMCID PMC13221074 · OpenAlex W7162800040
Open access: gold, a free copy (OpenAlex)
Status: code verified
Categories: genetics / omics (modality), human (organism), Parkinson's (population), cellular / molecular (subfield)
Methods: Statistics, Machine learning, Connectivity
MeSH: Brain Diseases*, Drug Repositioning*, Animals, Computational Biology, Genetic Predisposition to Disease, Genome-Wide Association Study, Graph Neural Networks, Humans, Parkinson Disease, Representation Machine Learning (* major topic)
Topic: Genetic Associations and Epidemiology (Genetics, Biochemistry, Genetics and Molecular Biology), according to OpenAlex
Funding: National Natural Science Foundation of China (82371482); Natural Science Foundation of Guangdong Province (2024A1515011363); National Key Research and Development Program of China (2023YFF1204902)
Citations: not cited yet (Europe PMC); 79 references in the paper

Abstract

Many genetic loci were identified as associated with neuropsychiatric disorders and neurodegenerative disorders by Genome-wide association studies (GWAS). How these loci impact these diseases is unclear. Advances in deep-learning approaches and multi-omics data have the potential to link GWAS findings with disease mechanisms. Here, we proposed the Multi-omics Graph Transformer Network (MOGT), a semi-supervised graph neural network that leverages graph representation learning to model biological networks derived from multi-omics data to predict disease-associated genes. MOGT outperforms the current approaches in disease gene prediction for two psychiatric disorders and three neurodegenerative/neurological diseases. High-risk genes (HRGs) for Parkinson’s disease (PD) predicted by MOGT were used to drug discovery by integrating with the CMAP database. Finally, 10 drugs were identified as potential candidates. Among them, the effect of drug UK-356618 was experimentally verified in a primary neuron model, showing that UK-356618 reversed the abnormal expression of PD-associated genes and improved the cell-level phenotypes of PD. Together, these results indicate that MOGT can be used to identify HRGs for brain disorders, and these predicted HRGs provide high-level insights into the mechanisms and treatments of brain disorders.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repository

Its files are read in the Code ↔ Paper reader above, with 4 matches between paragraphs and lines of code.

biomed-AI/MOGT

License: none: the authors keep all their rights
State: the link answers, verified on 28 September 2026
Evidence: files inventoried
Commit: 25ff432d66dc01849c119f0d4acc1d3802dfc7b2, 22 April 2026
Languages: Python (5), Jupyter (1)
Size: 20 files, 6 scripts
Software Heritage: not archived
Found in: “Data Availability”
Holds: README, environment (requirements.txt), 1 notebook
Not found: license file, CITATION.cff, tests, continuous integration, documentation
Tools: NumPy (3 files), pandas (3 files), PyTorch Geometric (3 files), PyTorch (3 files), Matplotlib (2 files), scikit-learn (2 files), Hugging Face Transformers (1 file)
Availability: 1 check, the latest on 28 September 2026: the link answers
  • 28 September 2026: the link answers
7 files
At the source: github.com/biomed-AI/MOGT

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 6 scripts, each with its path and the digest of its content;
  • 4 matches between paragraphs of the paper and lines of the code (method lexical-v1);
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

Datasets cited

Data Availability

All relevant data supporting the key findings of this study are available within the article and its Supplementary Material. In this study, we collected data from public databases. Genes related to brain disorders were obtained from four databases, Human Phenotype Ontology (HPO, https://hpo.jax.org/), Online Mendelian Inheritance in Man (OMIM, https://omim.org/), Disease Gene Network (DisGeNet, https://disgenet.com/), and GeneCards (https://www.genecards.org/). Highly expressed and lowly expressed gene sets in brain regions are from Human Protein Atlas (https://www.proteinatlas.org/). Gene-gene interactions are from BioGrid (https://thebiogrid.org/). The SNP-SNP interaction networks were obtained from a previous study [5]. Multi-omics for MOGT were described in the Supplementary Material. The source of bioinformatics analysis data is as given in the article. Code availability The source code to train the MOGT and reproduce the results is available in GitHub at https://github.com/biomed-AI/MOGT.

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 2, 28 September 2026

  • Funding: added National Natural Science Foundation of China: 82371482; Natural Science Foundation of Guangdong Province: 2024A1515011363; National Key Research and Development Program of China: 2023YFF1204902

Version 1, 28 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 5 authors, 10 MeSH terms, 79 references.

Cite

This paper

Li, J., Li, Y., Lin, S., Rao, J., & Zhao, H. (2026). A prototype-augmented graph representation learning framework for identifying brain disorder-associated genes and facilitating drug repurposing. PLoS computational biology, 22(5), e1014323. https://doi.org/10.1371/journal.pcbi.1014323

BibTeX

@article{li2026prototype,
author = {Li, Jiafang and Li, Yifei and Lin, Siying and Rao, Jiahua and Zhao, Huiying},
title = {{A prototype-augmented graph representation learning framework for identifying brain disorder-associated genes and facilitating drug repurposing}},
journal = {PLoS computational biology},
year = {2026},
month = may,
volume = {22},
number = {5},
pages = {e1014323},
publisher = {PLOS},
issn = {1553-734X},
doi = {10.1371/journal.pcbi.1014323},
url = {https://doi.org/10.1371/journal.pcbi.1014323},
pmid = {42213730},
pmcid = {PMC13221074}
}

RIS

TY - JOUR
AU - Li, Jiafang
AU - Li, Yifei
AU - Lin, Siying
AU - Rao, Jiahua
AU - Zhao, Huiying
TI - A prototype-augmented graph representation learning framework for identifying brain disorder-associated genes and facilitating drug repurposing
T2 - PLoS computational biology
J2 - PLoS Comput Biol
PY - 2026
DA - 2026/05/29
VL - 22
IS - 5
SP - e1014323
SN - 1553-734X
PB - PLOS
DO - 10.1371/journal.pcbi.1014323
UR - https://doi.org/10.1371/journal.pcbi.1014323
LA - en
ER -

CSL-JSON

{
"id": "10.1371/journal.pcbi.1014323",
"type": "article-journal",
"title": "A prototype-augmented graph representation learning framework for identifying brain disorder-associated genes and facilitating drug repurposing",
"container-title": "PLoS computational biology",
"author": [
{
"family": "Li",
"given": "Jiafang"
},
{
"family": "Li",
"given": "Yifei"
},
{
"family": "Lin",
"given": "Siying"
},
{
"family": "Rao",
"given": "Jiahua"
},
{
"family": "Zhao",
"given": "Huiying"
}
],
"container-title-short": "PLoS Comput Biol",
"volume": "22",
"issue": "5",
"page": "e1014323",
"DOI": "10.1371/journal.pcbi.1014323",
"PMID": "42213730",
"PMCID": "PMC13221074",
"ISSN": "1553-734X",
"publisher": "PLOS",
"URL": "https://doi.org/10.1371/journal.pcbi.1014323",
"language": "en",
"issued": {
"date-parts": [
[
2026,
5,
29
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1038/s41588-026-02646-3 [code]
Co-expression-based models improve eQTL predictions for transcriptome-wide association studies and highlight new schizophrenia-associated genes.
Journal: Nature genetics
In common: pandas, NumPy, genetics / omics, cellular / molecular, 6 references
[2] doi:10.3389/fnagi.2026.1806505 [code]
ALDH1A1-dopaminergic gene co-expression in human substantia nigra: meta-analysis of disease-associated correlation changes across seven independent Parkinson's disease datasets.
Journal: Frontiers in aging neuroscience
In common: pandas, Matplotlib, NumPy, NCBI GEO GSE7621, Parkinson's, cellular / molecular, 1 reference
[3] doi:10.3389/fnmol.2026.1844705 [code]
Risperidone regulates the expression of schizophrenia-related genes in the forebrain of adult male mice.
Journal: Frontiers in molecular neuroscience
In common: PyTorch, scikit-learn, pandas, 2 other tools, genetics / omics, cellular / molecular, 3 references
[4] doi:10.1038/s41467-026-73996-z [code]
Genetic architecture of white matter microstructure captured by unsupervised deep representation learning of fractional anisotropy maps.
Journal: Nature communications
In common: PyTorch, scikit-learn, pandas, 2 other tools, genetics / omics, cellular / molecular, 3 references
[5] doi:10.34133/csbj.0036 [code]
HYG-mol: An Interpretable Multimodal Hypergraph Framework for Molecular Property Prediction.
Journal: Computational and structural biotechnology journal
In common: PyTorch Geometric, Hugging Face Transformers, PyTorch, 4 other tools, cellular / molecular
[6] doi:10.1371/journal.pone.0351872 [code]
Decoding visual object recognition from EEG signals.
Journal: PloS one
In common: PyTorch Geometric, Hugging Face Transformers, PyTorch, 4 other tools
[7] doi:10.1038/s41467-026-71267-5 [code]
Human-like cognitive generalization for large models via mental representation-guided supervision.
Journal: Nature communications
In common: PyTorch Geometric, Hugging Face Transformers, PyTorch, 4 other tools
[8] doi:10.1038/s41467-026-75722-1 [code]
Single-nucleus analysis of the adult human olfactory epithelium uncovers shared neurogenesis programs with the brain.
Journal: Nature communications
In common: scikit-learn, pandas, Matplotlib, 1 other tool, genetics / omics, 4 references
[9] doi:10.1002/advs.77003 [code]
SemanticST: A Scalable Multi-Contextual Graph Learning Framework for Uncovering Spatial Niches and Robust Multi-Sample Integration in Spatial Transcriptomics.
Journal: Advanced science (Weinheim, Baden-Wurttemberg, Germany)
In common: PyTorch Geometric, PyTorch, scikit-learn, 3 other tools, genetics / omics, 1 reference
[10] doi:10.1371/journal.pone.0345854 [code]
Shedding light on neural learning to rank models for anticancer drug prioritization.
Journal: PloS one
In common: PyTorch Geometric, PyTorch, scikit-learn, 3 other tools, 1 reference

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.