Drug repurposing for PCOS by drug-target interaction mining

Planning brief for the Big Data mini research proposal. Team reference, September 2026. This is the plan and the facts behind it, not the submitted text: each member writes their own proposal in their own words.

1. The assignment in one table

SectionWhat it must containMarks
TitleDisease, data, method in one line3 (problem clarity and significance)
Biological problemUnmet need with a number, the specific gap, the promised output. Half a page.
Why big dataWhich Vs apply, each with a size. The step that breaks a laptop.2
DataReal public sources, size, format, licence, one example record each3 (data and approach)
ApproachNumbered pipeline, where Spark or distributed storage sits and why, a diagram
Expected outcomeConcrete deliverable and a numeric success measure2
ChallengesHardest technical part and one honest limitation

Length 2 to 3 pages. Hand-written or typed as instructed. Submitted through Google Classroom. The brief states that AI-generated text is detectable and receives no credit.

2. Why PCOS

Unmet need. Polycystic ovary syndrome affects roughly 1 in 10 women of reproductive age (Bozdag et al., 2016). No drug is approved specifically for PCOS. Treatment is symptomatic and borrowed: metformin (diabetes), combined oral contraceptives, spironolactone (hypertension), letrozole and clomiphene (fertility) (Teede et al., 2023). Insulin resistance, hyperandrogenism and chronic low-grade inflammation are the three biological axes (Escobar-Morreale, 2018).

The gap. The borrowed drugs were found one at a time by clinical intuition. There is no systematic, evidence-weighted ranking of all approved drugs against the PCOS gene set. That ranking is the promised output.

Why repurposing. A new drug takes 10 to 15 years and costs over a billion dollars (DiMasi et al., 2016); an approved drug already has safety and pharmacokinetic data, so a repurposing lead can move to a trial in a fraction of the time (Pushpakom et al., 2018).

The honest risk of PCOS as a choice. The disease gene set is small: about 19 to 20 GWAS loci (Day et al., 2018) and a few hundred associations in Open Targets, many with low evidence. The big-data argument therefore cannot rest on the disease side. It rests on the drug side (all bioactivity data), the network (the full human interactome) and, optionally, expression signatures.

3. Data sources

SourceRoleApproximate sizeOne recordLicence
ChEMBL v34/35
(Zdrazil et al., 2023)		Drug-target bioactivities2.4 M compounds, 20 M activities, 15 k targets. Dump ~4 GB compressed, ~25 GB loaded. Parquet on FTP.CHEMBL25 (aspirin), target CHEMBL2094253 (PTGS1), IC50 = 32 nM, pchembl 7.5CC BY-SA
DrugBank
(Knox et al., 2023)		Approved flag, indications16.5 k drugs (2.8 k approved small molecules), 20 k drug-target links. XML 1.5 GB.DB00331 metformin, approved, targets PRKAB1 (activator), indication: type 2 diabetesAcademic licence
BindingDB
(Liu et al., 2024)		Measured affinities, cross-check2.9 M measurements, 1.3 M compounds, 9 k targets. TSV ~1 GB.SMILES, UniProt P37231 (PPARG), Ki = 40 nM, PMIDCC BY 3.0
Open Targets Platform
(Ochoa et al., 2022); (Buniello et al., 2024)		PCOS-gene associations with evidence scores63 k targets, 28 k diseases, 18 M evidence strings. Parquet, ~50 GB total; associations slice ~2 GB.ENSG00000105329 (TGFB1), EFO_0000660 (PCOS), source ot_genetics_portal, score 0.42CC0
GWAS Catalog
(Sollis et al., 2022)		Genetic seed genes700 k associations. TSV ~300 MB.rs11031006, FSHB, polycystic ovary syndrome, p = 1e-12Open
STRING v12, human
(Szklarczyk et al., 2022)		Protein interaction network19.6 k proteins, 12 M scored links (1 M at score ≥ 700).9606.ENSP00000284981, 9606.ENSP00000262607, combined_score 912CC BY 4.0
LINCS L1000 (optional)
(Subramanian et al., 2017)		Drug-induced expression signatures3 M profiles, 34 k perturbagens, 978 landmark genes. Level 5 ~100+ GB.metformin, A549, 10 µM, 978 z-scoresOpen
GEO PCOS datasets (optional)
(Barrett et al., 2012)		Disease expression signatureGranulosa cell and adipose microarray/RNA-seq sets, tens to hundreds of samples each.GSE accession, sample, gene, log2 expressionOpen
UniProt / Ensembl mapping
(The UniProt Consortium, 2025); (Dyer et al., 2024)		Identifier bridgeHuman 20 k proteins; mapping files hundreds of MB.P37231 ↔ ENSG00000132170 ↔ 9606.ENSP00000287820 ↔ PPARGCC BY 4.0
ClinicalTrials.gov
(National Library of Medicine, n.d.)		Validation set500 k trials.NCT id, intervention metformin, condition PCOS, phase 3Open
Two corrections to the usual list. KEGG bulk download is licensed (Kanehisa et al., 2024), so use Reactome for pathways (Milacic et al., 2023). DisGeNET now restricts bulk access, so build the disease side on Open Targets and cite DisGeNET only as a cross-check (Piñero et al., 2019). Stating this shows the veracity check was done.

4. Why it is a big data problem, honestly

StepScaleLaptop?
Filter ChEMBL to approved drugs and join to targets20 M activity rows in, ~55 k rows outInput exceeds 8 GB RAM; needs a Spark join or a database. Output is small.
Build PCOS seed set from Open Targets + GWAS2 GB Parquet in, ~300 to 500 genes outSpark reads only the needed columns; result is small.
Network proximity for every approved drug on STRING2,800 drugs × 12 M edges, degree-matched random controls (1,000 per drug)No. ~2.8 million shortest-path computations. Embarrassingly parallel across drugs: Spark map over drug partitions with the graph broadcast.
Signature reversal against LINCS (optional)3 M profiles, 100+ GBNo. Chunked or distributed processing required.

The four Vs, with numbers. Volume: 20 M activities, 12 M edges, optionally 3 M expression profiles. Variety: bioactivity tables, evidence strings, a graph, expression matrices, trial records, each with its own identifiers. Veracity: predicted versus experimental interactions, evidence scores from 0 to 1, licence-restricted sources excluded. Velocity: ChEMBL and Open Targets release quarterly, so the pipeline must re-run, which is an argument for automation, not for streaming.

Why Spark rather than Hadoop MapReduce. Network propagation and proximity are iterative; Spark keeps intermediate state in memory (Zaharia et al., 2016), MapReduce writes to disk each iteration (Dean & Ghemawat, 2008). Open Targets already ships Parquet, Spark's native columnar format (Apache Software Foundation, n.d.).

5. Pipeline

ChEMBL + DrugBank
approved drugs → targets→ID mapping
UniProt ↔ Ensembl ↔ STRING→Drug target sets→Network proximity on STRING
Spark, per-drug map→Ranked candidates→Hold-out validation
known PCOS drugs
Open Targets + GWAS Catalog
PCOS seed genes with evidence scores→ feeds proximity step
  1. Ingest. Download ChEMBL (Parquet), DrugBank XML, Open Targets associations (Parquet), GWAS Catalog TSV, STRING human links. Store in HDFS or a local object store as Parquet. Tool: Spark.
  2. Approved drug set. ChEMBL molecules with max_phase = 4, cross-checked against DrugBank "approved". Output ~2,800 drugs.
  3. Drug target sets. ChEMBL activities with pchembl ≥ 6 (1 µM) on human single-protein targets, joined to the approved set. Spark join over 20 M rows. Output: drug → set of UniProt accessions.
  4. PCOS seed set. Open Targets associations for EFO_0000660 with overall score ≥ 0.1, plus GWAS Catalog mapped genes. Weight = evidence score. Output ~300 to 500 Ensembl genes.
  5. Identifier mapping. UniProt ↔ Ensembl ↔ STRING protein IDs. Log every unmapped ID; report the loss rate.
  6. Network proximity (Guney et al., 2016). For each drug: mean shortest-path distance from its targets to the seed set on STRING (score ≥ 400), compared with 1,000 degree-matched random target sets, giving a z-score. Spark map over drugs, graph as broadcast variable.
  7. Rank and annotate. Sort by z-score; attach for each drug its current indication, its closest PCOS genes and the evidence type behind them (the evidence path).
  8. Validate. Before step 6, hold out drugs already used or in trials for PCOS (metformin, spironolactone, letrozole, clomiphene, pioglitazone, liraglutide, statins, inositol where listed). Measure recall of the hold-out in the top 50 and AUROC against all other approved drugs, following the trial-based validation in Cheng et al., 2018.
  9. Optional second line of evidence. Signature reversal: PCOS granulosa or adipose signature from GEO versus LINCS drug signatures; report drugs that score on both lines.

6. Method options compared

MethodWhat it doesStrengthWeaknessReference
Direct overlap (baseline)Count of drug targets in the PCOS gene set, weighted by evidenceTrivially explainableOnly finds drugs on already-known genes—
Network proximity (chosen)Distance between targets and disease module versus randomExplainable, parallel, validated on many diseasesFavours well-studied proteins(Guney et al., 2016)
Random walk with restartDiffusion from seeds across the networkCaptures multi-step effectsIterative; parameter sensitive(Cowen et al., 2017)
Signature reversalDrugs whose expression signature opposes the disease signatureGenuinely big data; independent evidence lineCell lines are not ovary or adipose(Lamb et al., 2006); (Subramanian et al., 2017); (Sirota et al., 2011)
Knowledge-graph link predictionEmbedding model predicts drug-treats-disease edgesPowerfulHard to explain in 3 pages; leakage risk(Himmelstein et al., 2017)

7. Expected outcome and success measure

Deliverable. A ranked table of approved drugs with proximity z-score, current indication, closest PCOS genes and evidence type, plus a network figure for the top 10.

Success measure. At least 4 of the roughly 8 held-out drugs already used for PCOS appear in the top 50 of about 2,800, against under 0.2 expected by chance; AUROC above 0.7. If the hold-out is not recovered, the negative result still shows whether PCOS biology is reachable through the current target space, which is itself informative.

8. Challenges and limitations

9. Possible expansions

DirectionWhat it adds
LINCS signature reversal
(Subramanian et al., 2017); (Sirota et al., 2011)		Second, independent evidence line; the largest data axis (100+ GB).
Evidence-type ablation
(Buniello et al., 2024)		Tests whether genetic, expression or literature seeds drive the ranking; addresses veracity directly.
Side-effect filtering (SIDER, FAERS ~20 M reports)
(Kuhn et al., 2015); (U.S. Food and Drug Administration, n.d.)		Down-ranks candidates with adverse events relevant to women of reproductive age.
Tissue weighting (GTEx)
(The GTEx Consortium, 2020)		Weights targets by ovarian and adipose expression; cheap and biologically sensible.
Drug combinations
(Cheng et al., 2019)		~4 M pairs of approved drugs; combinatorial, so genuinely distributed.
Docking of top hits
(Corso et al., 2023)		Structural plausibility check for the top 10 candidates against their PCOS-relevant targets.
Literature mining (PubTator, 36 M abstracts)
(Wei et al., 2024)		Published co-mentions supporting or contradicting each candidate.
Cross-disease run (PCOS and type 2 diabetes)Shared insulin-resistance biology; turns the pipeline into a framework.
Knowledge-graph learning (Hetionet, PrimeKG)
(Himmelstein et al., 2017); (Chandak et al., 2023)		The machine-learning extension, done properly.
Real-world evidence
(Bycroft et al., 2018)		Incidence in cohorts of patients already on a candidate drug; access-controlled, future work.

10. Writing checklist

11. References (APA 7)

Journal entries were retrieved from the DOI registry (doi.org content negotiation) on 22 September 2026; the Cheng et al. (2018) entry was completed by hand after its DOI resolved. Verify each before submission.
  1. Apache Software Foundation. (n.d.). Apache Parquet documentation. https://parquet.apache.org/docs/
  2. Barrett, T., Wilhite, S. E., Ledoux, P., Evangelista, C., Kim, I. F., Tomashevsky, M., Marshall, K. A., Phillippy, K. H., Sherman, P. M., Holko, M., Yefanov, A., Lee, H., Zhang, N., Robertson, C. L., Serova, N., Davis, S., & Soboleva, A. (2012). NCBI GEO: archive for functional genomics data sets—update. Nucleic Acids Research, 41(D1), D991–D995. https://doi.org/10.1093/nar/gks1193
  3. Bozdag, G., Mumusoglu, S., Zengin, D., Karabulut, E., & Yildiz, B. O. (2016). The prevalence and phenotypic features of polycystic ovary syndrome: a systematic review and meta-analysis. Human Reproduction, 31(12), 2841–2855. https://doi.org/10.1093/humrep/dew218
  4. Buniello, A., Suveges, D., Cruz-Castillo, C., Llinares, M., Cornu, H., Lopez, I., Tsukanov, K., Roldán-Romero, J., Mehta, C., Fumis, L., McNeill, G., Hayhurst, J., Martinez Osorio, R., Barkhordari, E., Ferrer, J., Carmona, M., Uniyal, P., Falaguera, M., Rusina, P., et al. (2024). Open Targets Platform: facilitating therapeutic hypotheses building in drug discovery. Nucleic Acids Research, 53(D1), D1467–D1475. https://doi.org/10.1093/nar/gkae1128
  5. Bycroft, C., Freeman, C., Petkova, D., Band, G., Elliott, L. T., Sharp, K., Motyer, A., Vukcevic, D., Delaneau, O., O’Connell, J., Cortes, A., Welsh, S., Young, A., Effingham, M., McVean, G., Leslie, S., Allen, N., Donnelly, P., & Marchini, J. (2018). The UK Biobank resource with deep phenotyping and genomic data. Nature, 562(7726), 203–209. https://doi.org/10.1038/s41586-018-0579-z
  6. Chandak, P., Huang, K., & Zitnik, M. (2023). Building a knowledge graph to enable precision medicine. Scientific Data, 10(1). https://doi.org/10.1038/s41597-023-01960-3
  7. Cheng, F., Desai, R. J., Handy, D. E., Wang, R., Schneeweiss, S., Barabási, A.-L., & Loscalzo, J. (2018). Network-based approach to prediction and population-based validation of in silico drug repurposing. Nature Communications, 9(1), 2691. https://doi.org/10.1038/s41467-018-05116-5
  8. Cheng, F., Kovács, I. A., & Barabási, A.-L. (2019). Network-based prediction of drug combinations. Nature Communications, 10(1). https://doi.org/10.1038/s41467-019-09186-x
  9. Corso, G., Stärk, H., Jing, B., Barzilay, R., & Jaakkola, T. (2023). DiffDock: Diffusion steps, twists, and turns for molecular docking. International Conference on Learning Representations (ICLR 2023). https://arxiv.org/abs/2210.01189
  10. Cowen, L., Ideker, T., Raphael, B. J., & Sharan, R. (2017). Network propagation: a universal amplifier of genetic associations. Nature Reviews Genetics, 18(9), 551–562. https://doi.org/10.1038/nrg.2017.38
  11. Day, F., Karaderi, T., Jones, M. R., Meun, C., He, C., Drong, A., Kraft, P., Lin, N., Huang, H., Broer, L., Magi, R., Saxena, R., Laisk, T., Urbanek, M., Hayes, M. G., Thorleifsson, G., Fernandez-Tajes, J., Mahajan, A., Mullin, B. H., et al. (2018). Large-scale genome-wide meta-analysis of polycystic ovary syndrome suggests shared genetic architecture for different diagnosis criteria. PLOS Genetics, 14(12), e1007813. https://doi.org/10.1371/journal.pgen.1007813
  12. Dean, J., & Ghemawat, S. (2008). MapReduce. Communications of the ACM, 51(1), 107–113. https://doi.org/10.1145/1327452.1327492
  13. DiMasi, J. A., Grabowski, H. G., & Hansen, R. W. (2016). Innovation in the pharmaceutical industry: New estimates of R&D costs. Journal of Health Economics, 47, 20–33. https://doi.org/10.1016/j.jhealeco.2016.01.012
  14. Dyer, S. C., Austine-Orimoloye, O., Azov, A. G., Barba, M., Barnes, I., Barrera-Enriquez, V. P., Becker, A., Bennett, R., Beracochea, M., Berry, A., Bhai, J., Bhurji, S. K., Boddu, S., Branco Lins, P. R., Brooks, L., Ramaraju, S. B., Campbell, L. I., Martinez, M. C., Charkhchi, M., et al. (2024). Ensembl 2025. Nucleic Acids Research, 53(D1), D948–D957. https://doi.org/10.1093/nar/gkae1071
  15. Escobar-Morreale, H. F. (2018). Polycystic ovary syndrome: definition, aetiology, diagnosis and treatment. Nature Reviews Endocrinology, 14(5), 270–284. https://doi.org/10.1038/nrendo.2018.24
  16. Guney, E., Menche, J., Vidal, M., & Barábasi, A.-L. (2016). Network-based in silico drug efficacy screening. Nature Communications, 7(1). https://doi.org/10.1038/ncomms10331
  17. Himmelstein, D. S., Lizee, A., Hessler, C., Brueggeman, L., Chen, S. L., Hadley, D., Green, A., Khankhanian, P., & Baranzini, S. E. (2017). Systematic integration of biomedical knowledge prioritizes drugs for repurposing. eLife, 6. CLOCKSS. https://doi.org/10.7554/elife.26726
  18. Kanehisa, M., Furumichi, M., Sato, Y., Matsuura, Y., & Ishiguro-Watanabe, M. (2024). KEGG: biological systems database as a model of the real world. Nucleic Acids Research, 53(D1), D672–D677. https://doi.org/10.1093/nar/gkae909
  19. Knox, C., Wilson, M., Klinger, C., Franklin, M., Oler, E., Wilson, A., Pon, A., Cox, J., Chin, N. E., Strawbridge, S., Garcia-Patino, M., Kruger, R., Sivakumaran, A., Sanford, S., Doshi, R., Khetarpal, N., Fatokun, O., Doucet, D., Zubkowski, A., et al. (2023). DrugBank 6.0: the DrugBank Knowledgebase for 2024. Nucleic Acids Research, 52(D1), D1265–D1275. https://doi.org/10.1093/nar/gkad976
  20. Kuhn, M., Letunic, I., Jensen, L. J., & Bork, P. (2015). The SIDER database of drugs and side effects. Nucleic Acids Research, 44(D1), D1075–D1079. https://doi.org/10.1093/nar/gkv1075
  21. Lamb, J., Crawford, E. D., Peck, D., Modell, J. W., Blat, I. C., Wrobel, M. J., Lerner, J., Brunet, J.-P., Subramanian, A., Ross, K. N., Reich, M., Hieronymus, H., Wei, G., Armstrong, S. A., Haggarty, S. J., Clemons, P. A., Wei, R., Carr, S. A., Lander, E. S., & Golub, T. R. (2006). The Connectivity Map: Using Gene-Expression Signatures to Connect Small Molecules, Genes, and Disease. Science, 313(5795), 1929–1935. https://doi.org/10.1126/science.1132939
  22. Liu, T., Hwang, L., Burley, S., Nitsche, C., Southan, C., Walters, W., & Gilson, M. (2024). BindingDB in 2024: a FAIR knowledgebase of protein-small molecule binding data. Nucleic Acids Research, 53(D1), D1633–D1644. https://doi.org/10.1093/nar/gkae1075
  23. Milacic, M., Beavers, D., Conley, P., Gong, C., Gillespie, M., Griss, J., Haw, R., Jassal, B., Matthews, L., May, B., Petryszak, R., Ragueneau, E., Rothfels, K., Sevilla, C., Shamovsky, V., Stephan, R., Tiwari, K., Varusai, T., Weiser, J., et al. (2023). The Reactome Pathway Knowledgebase 2024. Nucleic Acids Research, 52(D1), D672–D678. https://doi.org/10.1093/nar/gkad1025
  24. National Library of Medicine. (n.d.). ClinicalTrials.gov [Database]. https://clinicaltrials.gov
  25. Ochoa, D., Hercules, A., Carmona, M., Suveges, D., Baker, J., Malangone, C., Lopez, I., Miranda, A., Cruz-Castillo, C., Fumis, L., Bernal-Llinares, M., Tsukanov, K., Cornu, H., Tsirigos, K., Razuvayevskaya, O., Buniello, A., Schwartzentruber, J., Karim, M., Ariano, B., et al. (2022). The next-generation Open Targets Platform: reimagined, redesigned, rebuilt. Nucleic Acids Research, 51(D1), D1353–D1359. https://doi.org/10.1093/nar/gkac1046
  26. Piñero, J., Ramírez-Anguita, J. M., Saüch-Pitarch, J., Ronzano, F., Centeno, E., Sanz, F., & Furlong, L. I. (2019). The DisGeNET knowledge platform for disease genomics: 2019 update. Nucleic Acids Research. https://doi.org/10.1093/nar/gkz1021
  27. Pushpakom, S., Iorio, F., Eyers, P. A., Escott, K. J., Hopper, S., Wells, A., Doig, A., Guilliams, T., Latimer, J., McNamee, C., Norris, A., Sanseau, P., Cavalla, D., & Pirmohamed, M. (2018). Drug repurposing: progress, challenges and recommendations. Nature Reviews Drug Discovery, 18(1), 41–58. https://doi.org/10.1038/nrd.2018.168
  28. Sirota, M., Dudley, J. T., Kim, J., Chiang, A. P., Morgan, A. A., Sweet-Cordero, A., Sage, J., & Butte, A. J. (2011). Discovery and Preclinical Validation of Drug Indications Using Compendia of Public Gene Expression Data. Science Translational Medicine, 3(96). https://doi.org/10.1126/scitranslmed.3001318
  29. Sollis, E., Mosaku, A., Abid, A., Buniello, A., Cerezo, M., Gil, L., Groza, T., Güneş, O., Hall, P., Hayhurst, J., Ibrahim, A., Ji, Y., John, S., Lewis, E., MacArthur, J. L., McMahon, A., Osumi-Sutherland, D., Panoutsopoulou, K., Pendlington, Z., et al. (2022). The NHGRI-EBI GWAS Catalog: knowledgebase and deposition resource. Nucleic Acids Research, 51(D1), D977–D985. https://doi.org/10.1093/nar/gkac1010
  30. Subramanian, A., Narayan, R., Corsello, S. M., Peck, D. D., Natoli, T. E., Lu, X., Gould, J., Davis, J. F., Tubelli, A. A., Asiedu, J. K., Lahr, D. L., Hirschman, J. E., Liu, Z., Donahue, M., Julian, B., Khan, M., Wadden, D., Smith, I. C., Lam, D., et al. (2017). A Next Generation Connectivity Map: L1000 Platform and the First 1,000,000 Profiles. Cell, 171(6), 1437–1452. https://doi.org/10.1016/j.cell.2017.10.049
  31. Szklarczyk, D., Kirsch, R., Koutrouli, M., Nastou, K., Mehryary, F., Hachilif, R., Gable, A. L., Fang, T., Doncheva, N., Pyysalo, S., Bork, P., Jensen, L., & von Mering, C. (2022). 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. https://doi.org/10.1093/nar/gkac1000
  32. Teede, H. J., Tay, C. T., Laven, J. J. E., Dokras, A., Moran, L. J., Piltonen, T. T., Costello, M. F., Boivin, J., Redman, L. M., Boyle, J. A., Norman, R. J., Mousa, A., Joham, A. E., Arlt, W., Azziz, R., Balen, A., Bedson, L., Berry, L., et al. (2023). Recommendations from the 2023 international evidence-based guideline for the assessment and management of polycystic ovary syndrome. European Journal of Endocrinology, 189(2), G43–G64. https://doi.org/10.1093/ejendo/lvad096
  33. The GTEx Consortium. (2020). The GTEx Consortium atlas of genetic regulatory effects across human tissues. Science, 369(6509), 1318–1330. https://doi.org/10.1126/science.aaz1776
  34. The UniProt Consortium. (2025). UniProt: The Universal Protein Knowledgebase in 2025. Nucleic Acids Research, 53(D1), D609–D617. https://doi.org/10.1093/nar/gkae1010
  35. U.S. Food and Drug Administration. (n.d.). FDA Adverse Event Reporting System (FAERS) public dashboard [Database]. https://fis.fda.gov/
  36. Wei, C.-H., Allot, A., Lai, P.-T., Leaman, R., Tian, S., Luo, L., Jin, Q., Wang, Z., Chen, Q., & Lu, Z. (2024). PubTator 3.0: an AI-powered literature resource for unlocking biomedical knowledge. Nucleic Acids Research, 52(W1), W540–W546. https://doi.org/10.1093/nar/gkae235
  37. Zaharia, M., Xin, R. S., Wendell, P., Das, T., Armbrust, M., Dave, A., Meng, X., Rosen, J., Venkataraman, S., Franklin, M. J., Ghodsi, A., Gonzalez, J., Shenker, S., & Stoica, I. (2016). Apache Spark. Communications of the ACM, 59(11), 56–65. https://doi.org/10.1145/2934664
  38. Zdrazil, B., Felix, E., Hunter, F., Manners, E. J., Blackshaw, J., Corbett, S., de Veij, M., Ioannidis, H., Lopez, D. M., Mosquera, J., Magarinos, M., Bosc, N., Arcila, R., Kizilören, T., Gaulton, A., Bento, A., Adasme, M., Monecke, P., Landrum, G., & Leach, A. (2023). The ChEMBL Database in 2023: a drug discovery platform spanning multiple bioactivity data types and time periods. Nucleic Acids Research, 52(D1), D1180–D1192. https://doi.org/10.1093/nar/gkad1004

Dataset sizes are approximate, from 2025 releases. Cite the release you download.