Pith. sign in

REVIEW 4 major objections 6 minor 49 references

VitaGraph: Building a Knowledge Graph for Biologically Relevant Learning Tasks

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read VitaGraph claims DRKG's drug-repurposing and PPI benchmark scores are inflated by data leakage, and provides a cleaned, feature-enriched knowledge graph for fair link-prediction benchmarking.

desk verdict Useful leakage audit and a plausible cleaned DRKG resource, but the benchmark evidence doesn't support the 'state-of-the-art' label. read the letter →

arxiv 2505.11185 v1 pith:PLYGP6DQ submitted 2025-05-16 cs.LG

Francesco Madeddu , Lucia Testa , Gianluca De Carlo , Michele Pieroni , Andrea Mastropietro , Aris Anagnostopoulos , Paolo Tieri , Sergio Barbarossa This is my paper · ORCID
classification cs.LG
keywords knowledgegraphdrugrepurposinglinkpredictiondataleakageprotein-proteininteractionsideeffectneuralnetworksbiomedicalbenchmark
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

VitaGraph is a biological knowledge graph built by cleaning, standardizing, and enriching the existing Drug Repurposing Knowledge Graph (DRKG). The paper's central claim is that DRKG's link-prediction results on drug repurposing and protein–protein interaction (PPI) tasks are inflated by data leakage between training and test splits, and that VitaGraph removes that leakage while adding expressive node features. If correct, this matters because machine-learning benchmarks on dirty integrated graphs can reward models for memorizing duplicate or synonymously labeled edges rather than for learning biology. The paper benchmarks VitaGraph on drug repurposing, PPI, and side-effect prediction, reporting performance comparable to or better than the un-cleaned graph and a leakage analysis showing 65.5% train/test overlap in PPI.

What carries the argument

The load-bearing mechanism is the leakage audit, not the model. The authors standardize relation labels and entity identifiers across the train/validation/test splits of each task, then count three kinds of overlap: exact duplicate triplets (including reversed edges), relation-level redundancy where the same semantic interaction appears under different relation IDs, and entity-level redundancy where the same entity appears under duplicated IDs. This audit quantifies how much of the test set is reachable from training information alone, and it is what exposes the 65.5% PPI and 18.7% drug-repurposing leakage that the paper argues invalidates direct comparison on the original DRKG.

What would settle it

Inspect VitaGraph's splits with an independent synonym dictionary: map every remaining relation label and entity ID to a canonical form using a source not used in the pipeline, then count duplicate triplets and reversed pairs across train and test. Any nonzero overlap after this second mapping would show the leakage-free guarantee is incomplete; for the side-effect task, the same check should return zero.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that the original DRKG cannot serve as a reliable benchmark for at least two of the three tasks: after standardizing relation labels and entity identifiers, 65.5% of PPI edges and about 18.7% of drug-repurposing edges in the test set duplicate training-set interactions under different names or IDs. VitaGraph is the authors' proposed remedy: a pipeline that filters malformed rows, maps synonymous relation labels to unified terms, unifies compound and disease identifiers, removes non-human and viral entries, de-duplicates exact and reversed triplets, drops compounds without SMILES, and then re-adds pathway and side-effect knowledge from curated sources while attaching Morgan fingerprints to compounds and one-hot functional profiles to genes. The benchmark results are presented not as a dramatic accuracy improvement over DRKG but as evidence that a fair comparison is possible only after cleaning, because a leakage-free version of DRKG would look structurally like VitaGraph without its enrichment.

Load-bearing premise

The reliability claim stands on the assumption that the mapping of synonymous relation labels and entity identifiers is complete, so that no semantically identical interaction survives under a different name after cleaning; if a synonym is missed, VitaGraph's own splits could retain the same leakage it criticizes in DRKG.

Editorial extensions

If this is right

  • Reported DRKG results on PPI and drug repurposing should be re-examined, since up to 65.5% of test interactions may already be present in training under synonym or duplicate labels.
  • VitaGraph provides a fairer benchmark: comparable or improved ranking metrics without the leakage hidden in the original graph.
  • Side-effect prediction is the one task where DRKG's original splits were not leaky, because the compound–side-effect edges come from a single source; this task can serve as a control in future knowledge-graph studies.
  • The cleaning pipeline, with its standardized relation labels and entity IDs, can be re-run or extended as new data sources arrive, making the benchmark maintainable.
  • Node features such as molecular fingerprints and gene functional profiles let graph models start from chemically and biologically meaningful representations rather than from random embeddings.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural extension is to apply the same three-way leakage audit to other integrated biological knowledge graphs; the audit is dataset-agnostic and could become a standard pre-benchmark check.
  • If the leakage rates generalize, published drug-repurposing rankings based on DRKG may shift once retrained on leakage-free splits; this is an editor's inference, not a claim in the paper.
  • The standardized relation-label mapping itself is a reusable artifact: future graph-construction efforts could adopt it as a dictionary for harmonizing heterogeneous biomedical vocabularies.
  • The leakage audit suggests a testable rule: any multi-source knowledge graph that merges duplicate entities and synonym relations should be audited for cross-split redundancy before its link-prediction scores are reported.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. This paper introduces VitaGraph, a biological knowledge graph derived from the Drug Repurposing Knowledge Graph (DRKG) through a multi-stage cleaning and enrichment pipeline. The pipeline removes formatting errors, non-human and virus-related entries, standardizes relation labels and entity identifiers, deduplicates edges, and enriches the graph with Reactome pathways, OnSIDES side-effect edges, Morgan fingerprints for compounds, and one-hot biological-process features for genes. The authors evaluate the resource with R-GCN and CompGCN on three link prediction tasks (drug repurposing, protein-protein interaction, side-effect prediction) across three dataset versions (original DRKG, cleaned VitaGraph without features, full VitaGraph). They further quantify data leakage in DRKG's random splits for the PPI and drug repurposing tasks and argue that VitaGraph is a leakage-free, state-of-the-art benchmark.

Significance. The principal contribution is a public, reproducible pipeline that produces a cleaner biological knowledge graph, together with a leakage analysis that quantifies train/test redundancy in DRKG; this is a genuinely useful cautionary result for the network medicine community. The dataset, code, and detailed appendices are valuable assets. However, the significance of the paper as a benchmark claim is currently limited by three gaps: the leakage-free property of VitaGraph itself is not measured; the benchmark results show mixed, overlapping performance with no significance tests; and the completeness of the relation-standardization mapping, on which the leakage analysis and de-duplication rest, is not fully documented. If these gaps are addressed, VitaGraph could become a credible benchmark resource.

major comments (4)
  1. [Section 4.3, Table 4] The leakage analysis is performed only on DRKG; no leakage ratios are reported for VitaGraph's own 70/10/20 splits. Since the paper's central reliability claim is that VitaGraph is a leakage-free benchmark (Sections 1 and 4.3), the authors should directly measure duplicate, relation-level, and entity-level redundancy in VitaGraph's splits using the same three checks, and report the numbers.
  2. [Section 2.2, Appendix A] The completeness of the relation-standardization mapping is a load-bearing assumption for both the leakage quantification in Table 4 and the de-duplication step in Section 2.5. However, Table 5 maps only 14 standardized labels for the original 107 edge types and does not list every original relation label; the text states that the complete mapping is in Appendix A, but the appendix presents only a source-presence matrix. The authors should provide an exhaustive mapping table (original label to standardized label) and justify groupings such as DrugBank Target, GNBR CbG, and IntAct Direct Interaction all mapped to CMP_BIND.
  3. [Tables 1-3] The experimental results do not support the "state-of-the-art platform" claim in the abstract. For instance, in Table 3 (PPI), VITAGRAPH with R-GCN achieves AUROC 0.859 ± 0.059 versus DRKG's 0.885 ± 0.032, and in Table 1 (drug repurposing), the VITAGRAPH R-GCN AUROC 0.887 ± 0.060 overlaps with DRKG's 0.914 ± 0.030. All comparisons show large overlapping standard deviations and no significance tests. Either temper the claims or provide proper statistical testing (e.g., paired tests across seeds, bootstrap confidence intervals on performance differences).
  4. [Section 3.5, Figure 1] The collapse of gene-pathway and gene-GO edges into binary one-hot node features discards relation-type information and edge semantics (e.g., the distinction between molecular function, biological process, and cellular component is only partially preserved, and any edge direction or weight is lost). The text claims this "reduced the graph's complexity while simultaneously enhancing the expressive capacity," but no ablation is provided. Please compare against a version of VitaGraph that retains GO and pathway nodes as edges, or otherwise justify that the feature collapse does not harm link prediction for the three tasks.
minor comments (6)
  1. [Section 3.6] The stated "a 21% decrease" for the reduction from 5,874,261 to 4,004,583 is incorrect; the actual reduction is about 31.8%. Please correct the percentage.
  2. [Section 4.1, Appendix B] Section 4.1 mentions only R-GCN and CompGCN, but Appendix B and Tables 8-16 include R-GAT; either add R-GAT to the architecture description and report its results in the main text, or remove it from the appendix.
  3. [Section 2.4] "118 redundant compound IDs" appears under the disease mapping paragraph; likely "disease IDs" is intended.
  4. [Table 4] Please state how many random splits were used to compute the means and standard deviations, and specify whether the splits were stratified by relation type.
  5. [Table 5] The abbreviations J_c, J_g, gene_OTHER_cmp, and gene_OTHER_gene are not defined; please add a legend.
  6. [Section 5] The limitation paragraph mentions only reliance on source data quality; it would be useful to also acknowledge the potential incompleteness of the relation-standardization mapping and the lack of a direct leakage measurement on VitaGraph.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular dependency: VitaGraph's construction is an external-data cleaning and feature-enrichment pipeline, and its leakage audit is an external-style check; score 0.

full rationale

The paper's central claims are that VitaGraph is a cleaned, standardized, feature-enriched knowledge graph and that DRKG's apparent benchmark results are inflated by leakage. These are data-assembly and empirical-audit claims, not derivations from fitted parameters or from self-citations. The relation-standardization mapping in Section 2.2 and Appendix A is load-bearing for the leakage analysis, and its completeness is an assumption that could be questioned, but a completeness weakness is a correctness risk, not a circular step: the paper does not define the leakage-free conclusion in terms of the mapping's output. The gene features in Section 3.5 are constructed by collapsing non-target DRKG edges (pathway, molecular function, biological process, and cellular component associations) into node feature vectors; these are not the target relations used in the benchmark tasks (drug–gene, gene–gene, and compound–side effect), so no prediction is equivalent to its input by construction. The leakage analysis in Section 4.3 is an external audit of DRKG splits using relation standardization, not a result forced by the model or by the dataset construction. No self-citation is load-bearing: the cited works by the authors [9, 10] are background examples of prior GDA methods, and the paper invokes no uniqueness theorem or ansatz from the authors' prior work. The paper also explicitly acknowledges its limitation of relying on source-data quality, which is a transparency statement rather than a circular justification. Overall, the derivation chain is self-contained with respect to the claims made.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

No new physical or modeled entities are introduced; the contribution is assembling and cleaning existing data. There are no free parameters in the scientific sense: model hyperparameters are tuned per task and reported in Appendix B, but they are not part of the dataset claim.

assumptions (4)
  • domain assumption The source databases (DRKG, Reactome, OnSIDES, UniChem, Disease Ontology) contain accurate and correctly mapped biological relationships.
    The entire knowledge graph inherits ground truth from these external resources; this is acknowledged as a limitation in Section 5, and no independent validation of edges is provided.
  • domain assumption Standardizing relation labels in Section 2.2 captures all semantic equivalences, so no residual leakage remains after cleaning.
    The leakage analysis in Section 4.3 relies on this mapping being complete; if a synonym is missed, test edges could still be inferred from train edges.
  • ad hoc to paper Collapsing gene-to-pathway and gene-to-GO edges into one-hot node features, as in Figure 1 and Section 3.5, loses no information needed for link prediction.
    This is a modeling choice specific to this paper; the claimed benefits of enrichment depend on it, but its sufficiency is not proven and the benchmark results are mixed.
  • domain assumption Morgan fingerprints of length 2,048 are a meaningful chemical representation for drug nodes.
    Standard in chemoinformatics, but the paper does not specify the radius or implementation, and it assumes these bits help machine-learning tasks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of VitaGraph: Building a Knowledge Graph for Biologically Relevant Learning Tasks." pith.science (2026). https://pith.science/paper/PLYGP6DQ

@misc{pith2026250511185,
  author       = {Pith},
  title        = {Pith review of: VitaGraph: Building a Knowledge Graph for Biologically Relevant Learning Tasks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PLYGP6DQ}},
  note         = {Machine review of arXiv:2505.11185}
}
read the original abstract

The intrinsic complexity of human biology presents ongoing challenges to scientific understanding. Researchers collaborate across disciplines to expand our knowledge of the biological interactions that define human life. AI methodologies have emerged as powerful tools across scientific domains, particularly in computational biology, where graph data structures effectively model biological entities such as protein-protein interaction (PPI) networks and gene functional networks. Those networks are used as datasets for paramount network medicine tasks, such as gene-disease association prediction, drug repurposing, and polypharmacy side effect studies. Reliable predictions from machine learning models require high-quality foundational data. In this work, we present a comprehensive multi-purpose biological knowledge graph constructed by integrating and refining multiple publicly available datasets. Building upon the Drug Repurposing Knowledge Graph (DRKG), we define a pipeline tasked with a) cleaning inconsistencies and redundancies present in DRKG, b) coalescing information from the main available public data sources, and c) enriching the graph nodes with expressive feature vectors such as molecular fingerprints and gene ontologies. Biologically and chemically relevant features improve the capacity of machine learning models to generate accurate and well-structured embedding spaces. The resulting resource represents a coherent and reliable biological knowledge graph that serves as a state-of-the-art platform to advance research in computational biology and precision medicine. Moreover, it offers the opportunity to benchmark graph-based machine learning and network medicine models on relevant tasks. We demonstrate the effectiveness of the proposed dataset by benchmarking it against the task of drug repurposing, PPI prediction, and side-effect prediction, modeled as link prediction problems.

Figures

Figures reproduced from arXiv: 2505.11185 by the authors.

Figure 1
Figure 1. Construction of the gene feature vector. A one-valued entry in the feature vector means that [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Comparison of the original DRKG and the new [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. VITAGRAPH creation pipeline. serve as strong benchmarks for assessing the utility and generalizability of knowledge graph-based representations in biomedical applications. To rigorously assess the impact of our dataset enhancements, we performed experiments across three distinct versions of the dataset: (1) the original DRKG dataset, which serves as a baseline, (2) a cleaned version of DRKG including the Reactome an… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 40 canonical work pages

  1. [1]

    From gpus to ai and quantum: three waves of acceleration in bioinformatics

    Bertil Schmidt and Andreas Hildebrandt. From gpus to ai and quantum: three waves of acceleration in bioinformatics. Drug Discovery Today, page 103990, 2024

    work page 2024

  2. [2]

    Discovering and linking public omics data sets using the omics discovery index

    Yasset Perez-Riverol, Mingze Bai, Felipe da Veiga Leprevost, Silvano Squizzato, Young Mi Park, Kenneth Haug, Adam J Carroll, Dylan Spalding, Justin Paschall, Mingxun Wang, et al. Discovering and linking public omics data sets using the omics discovery index. Nature biotechnology, 35(5):406–409, 2017

    work page 2017

  3. [3]

    Quantifying the impact of public omics data

    Yasset Perez-Riverol, Andrey Zorin, Gaurhari Dass, Manh-Tu Vu, Pan Xu, Mihai Glont, Juan Antonio Vizcaíno, Andrew F Jarnuczak, Robert Petryszak, Peipei Ping, et al. Quantifying the impact of public omics data. Nature communications, 10(1):3512, 2019

    work page 2019

  4. [4]

    Network medicine: a network- based approach to human disease

    Albert-László Barabási, Natali Gulbahce, and Joseph Loscalzo. Network medicine: a network- based approach to human disease. Nature reviews genetics, 12(1):56–68, 2011

    work page 2011

  5. [5]

    The biogrid database: A comprehensive biomedical resource of curated protein, genetic, and chemical interactions

    Rose Oughtred, Jennifer Rust, Christie Chang, Bobby-Joe Breitkreutz, Chris Stark, Andrew Willems, Lorrie Boucher, Genie Leung, Nadine Kolas, Frederick Zhang, et al. The biogrid database: A comprehensive biomedical resource of curated protein, genetic, and chemical interactions. Protein Science, 30(1):187–200, 2021

    work page 2021

  6. [6]

    The string database in 2023: protein–protein association networks and functional enrichment analyses for any sequenced genome of interest

    Damian Szklarczyk, Rebecca Kirsch, Mikaela Koutrouli, Katerina Nastou, Farrokh Mehryary, Radja Hachilif, Annika L Gable, Tao Fang, Nadezhda T Doncheva, Sampo Pyysalo, et al. The string database in 2023: protein–protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic acids research, 51(D1):D638–D646, 2023

    work page 2023

  7. [7]

    A reference map of the human binary protein interactome

    Katja Luck, Dae-Kyum Kim, Luke Lambourne, Kerstin Spirohn, Bridget E Begg, Wenting Bian, Ruth Brignall, Tiziana Cafarelli, Francisco J Campos-Laborie, Benoit Charloteaux, et al. A reference map of the human binary protein interactome. Nature, 580(7803):402–408, 2020

    work page 2020

  8. [8]

    The power of protein interaction networks for associating genes with diseases

    Saket Navlakha and Carl Kingsford. The power of protein interaction networks for associating genes with diseases. Bioinformatics, 26(8):1057–1063, 2010

    work page 2010

Show all 49 references
  1. [9]

    Niapu: network-informed adaptive positive-unlabeled learning for disease gene identification

    Paola Stolfi, Andrea Mastropietro, Giuseppe Pasculli, Paolo Tieri, and Davide Vergni. Niapu: network-informed adaptive positive-unlabeled learning for disease gene identification. Bioinfor- matics, 39(2):btac848, 2023

    2023

  2. [10]

    Xgdag: explainable gene–disease associations via graph neural networks

    Andrea Mastropietro, Gianluca De Carlo, and Aris Anagnostopoulos. Xgdag: explainable gene–disease associations via graph neural networks. Bioinformatics, 39(8):btad482, 2023

    2023

  3. [11]

    A survey on computational models for predicting protein–protein interactions

    Lun Hu, Xiaojuan Wang, Yu-An Huang, Pengwei Hu, and Zhu-Hong You. A survey on computational models for predicting protein–protein interactions. Briefings in bioinformatics, 22(5):bbab036, 2021

    2021

  4. [12]

    Gene regulatory network inference in the era of single-cell multi-omics

    Pau Badia-i Mompel, Lorna Wessels, Sophia Müller-Dott, Rémi Trimbour, Ricardo O Ramirez Flores, Ricard Argelaguet, and Julio Saez-Rodriguez. Gene regulatory network inference in the era of single-cell multi-omics. Nature Reviews Genetics, 24(11):739–754, 2023

    2023

  5. [13]

    Disgenet: a com- prehensive platform integrating information on human disease-associated genes and variants

    Janet Piñero, Àlex Bravo, Núria Queralt-Rosinach, Alba Gutiérrez-Sacristán, Jordi Deu-Pons, Emilio Centeno, Javier García-García, Ferran Sanz, and Laura I Furlong. Disgenet: a com- prehensive platform integrating information on human disease-associated genes and variants. Nucl...

    2016

  6. [14]

    The disgenet knowledge platform for disease genomics: 2019 update

    Janet Piñero, Juan Manuel Ramírez-Anguita, Josep Saüch-Pitarch, Francesco Ronzano, Emilio Centeno, Ferran Sanz, and Laura I Furlong. The disgenet knowledge platform for disease genomics: 2019 update. Nucleic acids research, 48(D1):D845–D855, 2020

    2019

  7. [15]

    Drugbank 6.0: the drugbank knowledgebase for 2024

    Craig Knox, Mike Wilson, Christen M Klinger, Mark Franklin, Eponine Oler, Alex Wilson, Allison Pon, Jordan Cox, Na Eun Chin, Seth A Strawbridge, et al. Drugbank 6.0: the drugbank knowledgebase for 2024. Nucleic acids research, 52(D1):D1265–D1275, 2024. 10

    2024

  8. [16]

    The comparative toxicogenomics database: update 2017

    Allan Peter Davis, Cynthia J Grondin, Robin J Johnson, Daniela Sciaky, Benjamin L King, Roy McMorran, Jolene Wiegers, Thomas C Wiegers, and Carolyn J Mattingly. The comparative toxicogenomics database: update 2017. Nucleic acids research, 45(D1):D972–D978, 2017

    2017

  9. [17]

    The sider database of drugs and side effects

    Michael Kuhn, Ivica Letunic, Lars Juhl Jensen, and Peer Bork. The sider database of drugs and side effects. Nucleic acids research, 44(D1):D1075–D1079, 2016

    2016

  10. [18]

    Data-driven prediction of drug effects and interactions

    Nicholas P Tatonetti, Patrick P Ye, Roxana Daneshjou, and Russ B Altman. Data-driven prediction of drug effects and interactions. Science translational medicine, 4(125):125ra31– 125ra31, 2012

    2012

  11. [19]

    Knowledge graphs for drug repurposing: a review of databases and methods

    Pablo Perdomo-Quinteiro and Alberto Belmonte-Hernández. Knowledge graphs for drug repurposing: a review of databases and methods. Briefings in Bioinformatics, 25(6):bbae461, 2024

    2024

  12. [20]

    An open source knowledge graph ecosystem for the life sciences

    Tiffany J Callahan, Ignacio J Tripodi, Adrianne L Stefanski, Luca Cappelletti, Sanya B Taneja, Jordan M Wyrwa, Elena Casiraghi, Nicolas A Matentzoglu, Justin Reese, Jonathan C Silverstein, et al. An open source knowledge graph ecosystem for the life sciences. Scientific Data, ...

    2024

  13. [21]

    Ioannidis, Xiang Song, Saurav Manchanda, Mufei Li, Xiaoqin Pan, Da Zheng, Xia Ning, Xiangxiang Zeng, and George Karypis

    Vassilis N. Ioannidis, Xiang Song, Saurav Manchanda, Mufei Li, Xiaoqin Pan, Da Zheng, Xia Ning, Xiangxiang Zeng, and George Karypis. Drkg - drug repurposing knowledge graph for covid-19. https://github.com/gnn4dr/DRKG/, 2020

    2020

  14. [22]

    Few-shot link prediction via graph neural networks for covid-19 drug-repurposing

    Vassilis N Ioannidis, Da Zheng, and George Karypis. Few-shot link prediction via graph neural networks for covid-19 drug-repurposing. arXiv preprint arXiv:2007.10261, 2020

    2007 arXiv

  15. [23]

    Molecular-evaluated and explainable drug repurpos- ing for covid-19 using ensemble knowledge graph embedding

    Md Kamrul Islam, Diego Amaya-Ramirez, Bernard Maigret, Marie-Dominique Devignes, Sabeur Aridhi, and Malika Smaïl-Tabbone. Molecular-evaluated and explainable drug repurpos- ing for covid-19 using ensemble knowledge graph embedding. Scientific Reports, 13(1):3643, 2023

    2023

  16. [24]

    Drugbank: a knowledgebase for drugs, drug actions and drug targets

    David S Wishart, Craig Knox, An Chi Guo, Dean Cheng, Savita Shrivastava, Dan Tzur, Bijaya Gautam, and Murtaza Hassanali. Drugbank: a knowledgebase for drugs, drug actions and drug targets. Nucleic acids research, 36(suppl_1):D901–D906, 2008

    2008

  17. [25]

    Systematic integration of biomedical knowledge prioritizes drugs for repurposing

    Daniel Scott Himmelstein, Antoine Lizee, Christine Hessler, Leo Brueggeman, Sabrina L Chen, Dexter Hadley, Ari Green, Pouya Khankhanian, and Sergio E Baranzini. Systematic integration of biomedical knowledge prioritizes drugs for repurposing. elife, 6:e26726, 2017

    2017

  18. [26]

    A global network of biomedical relationships derived from text

    Bethany Percha and Russ B Altman. A global network of biomedical relationships derived from text. Bioinformatics, 34(15):2614–2624, 2018

    2018

  19. [27]

    The intact database: efficient access to fine-grained molecular interaction data

    Noemi Del Toro, Anjali Shrivastava, Eliot Ragueneau, Birgit Meldal, Colin Combe, Elisabet Barrera, Livia Perfetto, Karyn How, Prashansa Ratan, Gautam Shirodkar, et al. The intact database: efficient access to fine-grained molecular interaction data. Nucleic acids research, 50 ...

    2022

  20. [28]

    Dgidb: mining the druggable genome

    Malachi Griffith, Obi L Griffith, Adam C Coffman, James V Weible, Josh F McMichael, Nicholas C Spies, James Koval, Indraniel Das, Matthew B Callaway, James M Eldred, et al. Dgidb: mining the druggable genome. Nature methods, 10(12):1209–1210, 2013

    2013

  21. [29]

    Dgidb 5.0: rebuilding the drug–gene interaction database for precision medicine and drug discovery platforms

    Matthew Cannon, James Stevenson, Kathryn Stahl, Rohit Basu, Adam Coffman, Susanna Kiwala, Joshua F McMichael, Kori Kuzma, Dorian Morrissey, Kelsy Cotto, et al. Dgidb 5.0: rebuilding the drug–gene interaction database for precision medicine and drug discovery platforms. Nucleic...

    2024

  22. [30]

    Gene ontology: tool for the unification of biology

    Michael Ashburner, Catherine A Ball, Judith A Blake, David Botstein, Heather Butler, J Michael Cherry, Allan P Davis, Kara Dolinski, Selina S Dwight, Janan T Eppig, et al. Gene ontology: tool for the unification of biology. Nature genetics, 25(1):25–29, 2000

    2000

  23. [31]

    Unichem: a unified chemical structure cross-referencing and identifier tracking system

    Jon Chambers, Mark Davies, Anna Gaulton, Anne Hersey, Sameer Velankar, Robert Petryszak, Janna Hastings, Louisa Bellis, Shaun McGlinchey, and John P Overington. Unichem: a unified chemical structure cross-referencing and identifier tracking system. Journal of cheminformatics, ...

    2013

  24. [32]

    Chembl: a large-scale bioactivity database for drug discovery

    Anna Gaulton, Louisa J Bellis, A Patricia Bento, Jon Chambers, Mark Davies, Anne Hersey, Yvonne Light, Shaun McGlinchey, David Michalovich, Bissan Al-Lazikani, et al. Chembl: a large-scale bioactivity database for drug discovery. Nucleic acids research, 40(D1):D1100– D1107, 2012

    2012

  25. [33]

    The chembl database in 2023: a drug discovery platform spanning multiple bioactivity data types and time periods

    Barbara Zdrazil, Eloy Felix, Fiona Hunter, Emma J Manners, James Blackshaw, Sybilla Corbett, Marleen de Veij, Harris Ioannidis, David Mendez Lopez, Juan F Mosquera, et al. The chembl database in 2023: a drug discovery platform spanning multiple bioactivity data types and time ...

    2023

  26. [34]

    Chebi: a database and ontology for chemical entities of biological interest

    Kirill Degtyarenko, Paula De Matos, Marcus Ennis, Janna Hastings, Martin Zbinden, Alan McNaught, Rafael Alcántara, Michael Darsow, Mickaël Guedj, and Michael Ashburner. Chebi: a database and ontology for chemical entities of biological interest. Nucleic acids research, 36 (sup...

    2007

  27. [35]

    Chebi in 2016: Improved services and an expanding collection of metabolites

    Janna Hastings, Gareth Owen, Adriano Dekker, Marcus Ennis, Namrata Kale, Venkatesh Muthukrishnan, Steve Turner, Neil Swainston, Pedro Mendes, and Christoph Steinbeck. Chebi in 2016: Improved services and an expanding collection of metabolites. Nucleic acids research, 44(D1):D1...

    2016

  28. [36]

    Pubchem: integrated platform of small molecules and biological activities

    Evan E Bolton, Yanli Wang, Paul A Thiessen, and Stephen H Bryant. Pubchem: integrated platform of small molecules and biological activities. In Annual reports in computational chemistry, volume 4, pages 217–241. Elsevier, 2008

    2008

  29. [37]

    Pubchem as a public resource for drug discovery

    Qingliang Li, Tiejun Cheng, Yanli Wang, and Stephen H Bryant. Pubchem as a public resource for drug discovery. Drug discovery today, 15(23-24):1052–1057, 2010

    2010

  30. [38]

    Pubchem 2025 update

    Sunghwan Kim, Jie Chen, Tiejun Cheng, Asta Gindulyte, Jia He, Siqian He, Qingliang Li, Benjamin A Shoemaker, Paul A Thiessen, Bo Yu, et al. Pubchem 2025 update. Nucleic Acids Research, 53(D1):D1516–D1525, 2025

    2025

  31. [39]

    Human disease ontology 2018 update: classification, content and workflow expansion

    Lynn M Schriml, Elvira Mitraka, James Munro, Becky Tauber, Mike Schor, Lance Nickle, Victor Felix, Linda Jeng, Cynthia Bearer, Richard Lichenstein, et al. Human disease ontology 2018 update: classification, content and workflow expansion. Nucleic acids research, 47(D1): D955–D...

    2018

  32. [40]

    The human disease ontology 2022 update

    Lynn M Schriml, James B Munro, Mike Schor, Dustin Olley, Carrie McCracken, Victor Felix, J Allen Baron, Rebecca Jackson, Susan M Bello, Cynthia Bearer, et al. The human disease ontology 2022 update. Nucleic acids research, 50(D1):D1255–D1261, 2022

    2022

  33. [41]

    Reactome: a knowledge base of biologic pathways and processes

    Imre Vastrik, Peter D’Eustachio, Esther Schmidt, Geeta Joshi-Tope, Gopal Gopinath, David Croft, Bernard de Bono, Marc Gillespie, Bijay Jassal, Suzanna Lewis, et al. Reactome: a knowledge base of biologic pathways and processes. Genome biology, 8:1–13, 2007

    2007

  34. [42]

    The reactome pathway knowledgebase 2024

    Marija Milacic, Deidre Beavers, Patrick Conley, Chuqiao Gong, Marc Gillespie, Johannes Griss, Robin Haw, Bijay Jassal, Lisa Matthews, Bruce May, et al. The reactome pathway knowledgebase 2024. Nucleic acids research, 52(D1):D672–D678, 2024

    2024

  35. [43]

    Onsides database: Extracting adverse drug events from drug labels using natural language processing models

    Yutaro Tanaka, Hsin Yi Chen, Pietro Belloni, Undina Gisladottir, Jenna Kefeli, Jason Patterson, Apoorva Srinivasan, Michael Zietz, Gaurav Sirdeshmukh, Jacob Berkowitz, et al. Onsides database: Extracting adverse drug events from drug labels using natural language processing mo...

    2025

  36. [44]

    Smiles, a chemical language and information system

    David Weininger. Smiles, a chemical language and information system. 1. introduction to methodology and encoding rules. Journal of Chemical Information and Computer Sciences, 28 (1):31–36, 1988

    1988

  37. [45]

    Harry L. Morgan. The generation of a unique machine description for chemical structures—a technique developed at chemical abstracts service. Journal of Chemical Documentation, 5(2): 107–113, 1965

    1965

  38. [46]

    Extended-connectivity fingerprints

    David Rogers and Mathew Hahn. Extended-connectivity fingerprints. Journal of Chemical Information and Modeling, 50(5):742–754, 2010. 12

    2010

  39. [47]

    One molecular fingerprint to rule them all: drugs, biomolecules, and the metabolome

    Alice Capecchi, Daniel Probst, and Jean-Louis Reymond. One molecular fingerprint to rule them all: drugs, biomolecules, and the metabolome. Journal of cheminformatics, 12:1–15, 2020

    2020

  40. [48]

    Modeling relational data with graph convolutional networks

    Michael Schlichtkrull, Thomas N Kipf, Peter Bloem, Rianne Van Den Berg, Ivan Titov, and Max Welling. Modeling relational data with graph convolutional networks. In The semantic web: 15th international conference, ESWC 2018, Heraklion, Crete, Greece, June 3–7, 2018, proceedings...

    2018

  41. [49]

    Activator

    Shikhar Vashishth, Soumya Sanyal, Vikram Nitin, and Partha Talukdar. Composition-based multi-relational graph convolutional networks. arXiv preprint arXiv:1911.03082, 2019. 13 A Details on V ITAGRAPH structure Table 5 shows the mapping between the interaction types in VITAGRAP...

    1911 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.