hssling/parkinsons-evidence-to-discovery-prioritisation

Parkinson's Disease Evidence-to-Discovery Prioritisation Dataset

This Hugging Face dataset package contains processed research assets from an AI-assisted evidence synthesis and computational validation project on Parkinson's disease (PD) prevention and disease-modifying therapeutic strategy prioritisation.

Dataset Summary

The dataset integrates:

• —evidence-priority scores for PD prevention and disease-modification candidates;
• —pathway-to-intervention framework;
• —individual and public-health prevention measures;
• —ClinicalTrials.gov mining outputs;
• —GSE72267 blood transcriptomic differential-expression outputs;
• —GO biological-process enrichment;
• —exploratory ReactomePA enrichment;
• —drug-repurposing candidate rankings;
• —multi-omics expansion inventory and pathway-recurrence gap map;
• —public NCBI GEO/GDS discovery results for candidate brain, blood, proteomics, metabolomics, methylation and single-cell datasets;
• —genetic causal-triangulation and variant-to-pathway scoring matrices with Open Targets and GWAS Catalog API outputs;
• —drug-discovery deepening outputs covering LINCS access status, ChEMBL selectivity, ADMET/BBB triage, docking readiness and ClinicalTrials.gov trial gaps;
• —experimental-validation work packages with assays, controls and go/no-go criteria;
• —publication figures, scripts, tables, manuscript files, reference audits, and reproducibility logs.

The dataset is intended for research, education, validation, reproducibility, and hypothesis generation. It does not claim that a cure has been found.

Dataset Structure

data/
  processed/
    candidate_interventions_ranked.csv
    pathway_intervention_framework.csv
    individual_prevention_measures.csv
    evidence_map.csv
  clinical_trials/
    clinical_trials_landscape.csv
  omics/
    GSE72267_limma_DEG_all.csv
    GSE72267_DEG_summary.csv
    GSE72267_GO_BP_enrichment_top500.csv
    GSE72267_Reactome_enrichment_top500_exploratory.csv
  drug_repurposing/
    drug_repurposing_candidates.csv
  omics_expansion/
    multi_omics_dataset_inventory.csv
    multi_tissue_pathway_recurrence.csv
    omics_modality_gap_map.csv
    public_multiomics_dataset_discovery.csv
  genetics/
    genetic_causal_triangulation_matrix.csv
    variant_to_pathway_scoring.csv
    opentargets_pd_association_scores.csv
    gwas_catalog_pd_gene_summary.csv
    gwas_catalog_pd_target_overlap.csv
  drug_discovery_deepening/
    drug_discovery_deepening_matrix.csv
    docking_readiness.csv
    clinical_trial_gap_map.csv
    chembl_compound_selectivity_summary.csv
    clinicaltrials_public_api_gap_query.csv
figures/
scripts/
manuscript/
tables/
reproducibility/
validation_work_packages/
references_audit.csv

Public API modules have been executed where possible. Formal colocalisation, Mendelian randomisation, LINCS/Connectivity Map signature reversal, patent/freedom-to-operate review and decision-grade docking remain labelled as blocked where required inputs, credentials or specialist workflows are not available.

Current Key Findings

The highest-priority candidates in the current scoring framework were:

1. 1.pesticide-exposure reduction;
2. 2.structured aerobic/resistance exercise;
3. 3.GLP-1 receptor pathway strategies;
4. 4.Mediterranean/MIND-style dietary pattern;
5. 5.LRRK2 inhibition;
6. 6.GBA/lysosomal modulation;
7. 7.head-injury prevention;
8. 8.air-pollution reduction.

GSE72267 limma analysis tested 22,277 probes across 40 PD and 19 control blood samples. No probes met the prespecified FDR < 0.05 and absolute log2 fold-change >= 0.25 threshold. GO and Reactome outputs are therefore exploratory and hypothesis-generating.

Repository and CI

The GitHub repository with version-controlled evidence assets, scripts, manuscript files, reproducibility checks, and CI validation is available at:

https://github.com/hssling/parkinsons-evidence-to-discovery-prioritisation

The repository CI runs a lightweight validation script that checks required release assets, intervention-ranking integrity, cautionary clinical-claim language, and NMJI reference ordering.

Intended Uses

• —Validate or modify the evidence scoring rubric.
• —Reanalyse pathway-intervention rankings.
• —Extend drug-repurposing prioritisation.
• —Reproduce GSE72267 omics analysis.
• —Teach biomedical evidence synthesis workflows.
• —Generate hypotheses for future PD prevention or disease-modification studies.

Out-of-Scope Uses

This dataset must not be used as:

• —a diagnostic tool;
• —clinical decision support;
• —a treatment recommendation system;
• —evidence that any intervention cures PD;
• —evidence that investigational or repurposed drugs should be used outside approved indications or clinical trials.

Data Sources

Upstream resources include public literature records, ClinicalTrials.gov, GEO accession GSE72267, GO, Reactome, and manually curated source-backed references. See references_audit.csv and reproducibility/methods_log.md.

Licensing

Generated processed data, code, figures, and documentation are released under CC BY 4.0. Users must respect terms and citation requirements of upstream resources.

Citation

Please cite this dataset and the upstream resources listed in references_audit.csv.