OSCR

ALDH1A1-dopaminergic gene co-expression in human substantia nigra: meta-analysis of disease-associated correlation changes across seven independent Parkinson's disease datasets.

Code ↔ Paper

12 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 12 matches
  1. [1] § Materials and methods › Data acquisition and processing ↔ 01_download_and_correlate.py, lines 1–54 · score 0.94 · multiple brain regions, HG U133A, HG U133B, biological sample, substantia nigra, GPL97
  2. [2] § Materials and methods › Cell type enrichment analysis ↔ config.py, lines 66–110 · score 0.80 · GABAergic, CALB1, astrocytes, endothelial, glutamatergic, oligodendrocytes
  3. [3] § Materials and methods › Meta-analysis ↔ 02_meta_analysis.py, lines 51–115 · score 0.75 · DerSimonian, transformed correlations, standard error, Cochran, SE, Laird
  4. [4] § Results › ALDH1A1 correlations are attenuated in Parkinson’s disease ↔ 02_meta_analysis.py, lines 155–287 · score 0.73 · DA pair, pooled correlations, ALDH1A1 DDC, ALDH1A1 DA, ALDH1A1 SLC18A2, gene pair
  5. [5] § Materials and methods › Negative control gene pairs ↔ config.py, lines 14–45 · score 0.71 · RPL13A, B2M, HPRT1, RPS18, PPIA, UBC
  6. [6] § Materials and methods › Cell type enrichment analysis ↔ 03_cell_type_enrichment.py, lines 1–35 · score 0.67 · scored expression, enrichment scores, regression, squares, deconvolution, NNLS
  7. [7] § Materials and methods › Cell type enrichment analysis ↔ 03_cell_type_enrichment.py, lines 1–35 · score 0.67 · enrichment scoring, SLC6A3, SLC18A2, circularity, target genes, signatures
  8. [8] § Materials and methods › Literature search and dataset selection ↔ 01_download_and_correlate.py, lines 1–54 · score 0.66 · brain regions, substantia nigra, profiling, microarray, GEO, GSE7621
  9. [9] § Materials and methods › Cell type enrichment analysis ↔ 03_cell_type_enrichment.py, lines 220–282 · score 0.64 · shuffled disease, common nodes, gene pairs sharing, permutation, enrichment, Selectivity
  10. [10] § Results › Selectivity of ALDH1A1 correlation attenuation ↔ 03_cell_type_enrichment.py, lines 220–282 · score 0.56 · shuffled disease, common node, gene pairs sharing, permutation, selectivity, correlation
  11. [11] § Results › Cell type enrichment analysis ↔ 03_cell_type_enrichment.py, lines 325–368 · score 0.56 · enrichment scores, adjusted correlations, PD samples, Cohen, vulnerable, raw
  12. [12] § Results › SNCA-containing pairs show intermediate to large attenuation ↔ 04_sensitivity_analysis.py, lines 1–32 · score 0.53 · selectivity comparison, SLC6A3, SLC18A2, ALDH1A1, DDC, PD

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

Python · 478 lines · 17 KB · MIT · 5 matches

  1. #!/usr/bin/env python3
  2. """
  3. 03_cell_type_enrichment.py
  4. ===========================
  5. Reference-based cell type enrichment analysis using marker genes from
  6. Kamath et al. (2022) single-nucleus RNA-seq of human substantia nigra.
  8. Key design features:
  9. - All six target genes (ALDH1A1, TH, DDC, SLC18A2, SLC6A3, SNCA)
  10. are EXCLUDED from all cell type signatures (zero circularity)
  11. - Enrichment scoring: mean z-scored expression of marker genes per sample
  12. - NNLS regression: non-negative least squares for proportion estimation
  13. - Partial correlations controlling for DA_Vulnerable enrichment
  14. - Permutation testing (sample-level label shuffling) for selectivity
  16. Note: GSE7621 is excluded from deconvolution (6 datasets analyzed).
  18. Reads: GEO datasets (re-downloads or loads from cache)
  19. Produces:
  20. results/deconvolution_results.csv
  21. results/raw_vs_adjusted_correlations.csv
  22. results/selectivity_analysis.csv
  23. results/permutation_test.csv
  24. """
  26. import os
  27. import sys
  28. import warnings
  29. import numpy as np
  30. import pandas as pd
  31. from scipy import stats
  32. from scipy.optimize import nnls
  33. from itertools import combinations
  35. warnings.filterwarnings('ignore')
  37. # Import shared configuration
  38. sys.path.insert(0, os.path.dirname(__file__))
  39. from config import (
  40. TARGET_GENES, HOUSEKEEPING_PAIRS, GENE_PAIR_CATEGORIES,
  41. DATASETS, DECONV_DATASETS, VALIDATED_DATASETS,
  42. CELL_TYPE_SIGNATURES, get_results_dir, get_category
  43. )
  45. # Import dataset processors from script 01 (can't import directly due to numeric prefix)
  46. import importlib.util
  47. spec = importlib.util.spec_from_file_location(
  48. "download", os.path.join(os.path.dirname(__file__), "01_download_and_correlate.py"))
  49. download_module = importlib.util.module_from_spec(spec)
  50. spec.loader.exec_module(download_module)
  52. download_series_matrix = download_module.download_series_matrix
  53. process_GSE8397 = download_module.process_GSE8397
  54. process_GSE49036 = download_module.process_GSE49036
  55. process_standard_dataset = download_module.process_standard_dataset
  57. RESULTS_DIR = get_results_dir()
  59. # Excluded target genes (for signature verification)
  60. EXCLUDED_GENES = set(TARGET_GENES)
  62. # Verify no target gene contamination
  63. for ct, genes in CELL_TYPE_SIGNATURES.items():
  64. overlap = set(genes) & EXCLUDED_GENES
  65. assert len(overlap) == 0, f"Target gene overlap in {ct}: {overlap}"
  68. # ---------------------------------------------------------------------------
  69. # Enrichment Scoring
  70. # ---------------------------------------------------------------------------
  71. def compute_enrichment_scores(expr_df, signatures):
  72. """Compute cell type enrichment scores as mean z-scored marker expression.
  74. Parameters
  75. ----------
  76. expr_df : DataFrame, gene × sample
  77. signatures : dict, cell_type → list of marker genes
  79. Returns
  80. -------
  81. DataFrame, sample × cell_type enrichment scores
  82. """
  83. # Z-score each gene across all samples
  84. z_df = expr_df.apply(lambda x: (x - x.mean()) / x.std(), axis=1)
  85. z_df = z_df.replace([np.inf, -np.inf], np.nan)
  87. scores = {}
  88. markers_used = {}
  89. for ct, markers in signatures.items():
  90. available = [g for g in markers if g in z_df.index]
  91. markers_used[ct] = (len(available), len(markers))
  92. if len(available) >= 3:
  93. scores[ct] = z_df.loc[available].mean(axis=0)
  94. else:
  95. scores[ct] = pd.Series(np.nan, index=expr_df.columns)
  97. return pd.DataFrame(scores), markers_used
  100. def compute_nnls_proportions(expr_df, signatures):
  101. """Estimate cell type proportions using non-negative least squares.
  103. This is mathematically equivalent to the core CIBERSORTx algorithm.
  104. """
  105. # Build reference matrix
  106. all_markers = []
  107. ct_labels = []
  108. for ct, markers in signatures.items():
  109. available = [g for g in markers if g in expr_df.index]
  110. all_markers.extend(available)
  111. ct_labels.extend([ct] * len(available))
  113. if not all_markers:
  114. return None
  116. # Create binary reference matrix
  117. ref = pd.DataFrame(0.0, index=all_markers, columns=list(signatures.keys()))
  118. for gene, ct in zip(all_markers, ct_labels):
  119. ref.loc[gene, ct] = 1.0
  121. # Normalize reference columns
  122. ref = ref / ref.sum(axis=0)
  124. # NNLS for each sample
  125. proportions = {}
  126. for sample in expr_df.columns:
  127. y = expr_df.loc[ref.index, sample].values.astype(float)
  128. mask = ~np.isnan(y)
  129. if mask.sum() < 5:
  130. continue
  131. A = ref.values[mask]
  132. b = y[mask]
  133. x, _ = nnls(A, b)
  134. x = x / x.sum() if x.sum() > 0 else x
  135. proportions[sample] = dict(zip(ref.columns, x))
  137. return pd.DataFrame(proportions).T
  140. # ---------------------------------------------------------------------------
  141. # Partial Correlation
  142. # ---------------------------------------------------------------------------
  143. def partial_corr(x, y, z):
  144. """Partial Pearson correlation between x and y, controlling for z.
  146. Parameters: x, y, z are 1D arrays of equal length.
  147. Returns: partial correlation coefficient.
  148. """
  149. mask = ~(np.isnan(x) | np.isnan(y) | np.isnan(z))
  150. x, y, z = x[mask], y[mask], z[mask]
  152. if len(x) < 5:
  153. return np.nan
  155. # Residualize x and y on z
  156. _, _, r_xz, _, _ = stats.linregress(z, x)
  157. _, _, r_yz, _, _ = stats.linregress(z, y)
  159. # Actually compute residuals
  160. slope_xz, int_xz = np.polyfit(z, x, 1)
  161. slope_yz, int_yz = np.polyfit(z, y, 1)
  163. res_x = x - (slope_xz * z + int_xz)
  164. res_y = y - (slope_yz * z + int_yz)
  166. if np.std(res_x) == 0 or np.std(res_y) == 0:
  167. return np.nan
  169. r, _ = stats.pearsonr(res_x, res_y)
  170. return r
  173. # ---------------------------------------------------------------------------
  174. # Selectivity Analysis
  175. # ---------------------------------------------------------------------------
  176. def compute_selectivity(corr_records, datasets=None):
  177. """Compute raw or adjusted selectivity between ALDH1A1-DA and DA-DA pairs.
  179. Returns dict with mean Δr by category, selectivity, and statistical tests.
  180. """
  181. df = pd.DataFrame(corr_records)
  182. if datasets is not None:
  183. df = df[df['dataset'].isin(datasets)]
  185. aldh1a1 = df[df['category'] == 'ALDH1A1-DA']['delta_r'].dropna()
  186. dada = df[df['category'] == 'DA-DA']['delta_r'].dropna()
  188. if len(aldh1a1) < 2 or len(dada) < 2:
  189. return None
  191. mean_a = aldh1a1.mean()
  192. mean_d = dada.mean()
  193. selectivity = mean_a - mean_d
  195. # Welch's t-test
  196. t_stat, t_p = stats.ttest_ind(aldh1a1, dada, equal_var=False)
  198. # Mann-Whitney U
  199. u_stat, u_p = stats.mannwhitneyu(aldh1a1, dada, alternative='two-sided')
  201. # Cohen's d
  202. pooled_std = np.sqrt((aldh1a1.var() * (len(aldh1a1) - 1) + dada.var() * (len(dada) - 1)) /
  203. (len(aldh1a1) + len(dada) - 2))
  204. d = (mean_a - mean_d) / pooled_std if pooled_std > 0 else np.nan
  206. return {
  207. 'ALDH1A1_mean_dr': mean_a,
  208. 'DADA_mean_dr': mean_d,
  209. 'selectivity': selectivity,
  210. 't_stat': t_stat,
  211. 't_p': t_p,
  212. 'U_stat': u_stat,
  213. 'U_p': u_p,
  214. 'cohens_d': d,
  215. 'n_ALDH1A1': len(aldh1a1),
  216. 'n_DADA': len(dada),
  217. }
  220. def permutation_test(all_data, n_perm=5000, datasets=None):
  221. """Permutation test for selectivity by shuffling disease labels within datasets.
  223. Preserves the dependency structure among gene pairs sharing common nodes.
  224. """
  225. if datasets is None:
  226. datasets = all_data['dataset'].unique()
  228. # Observed selectivity
  229. obs = compute_selectivity(
  230. [r for r in all_data if r.get('dataset') in datasets],
  231. datasets
  232. )
  233. if obs is None:
  234. return None
  235. obs_sel = obs['selectivity']
  237. print(f" Observed selectivity: {obs_sel:.4f}")
  238. print(f" Running {n_perm} permutations...")
  240. perm_sels = []
  241. for i in range(n_perm):
  242. if (i + 1) % 1000 == 0:
  243. print(f" Completed {i+1}/{n_perm}...")
  245. # Shuffle disease labels within each dataset
  246. perm_records = []
  247. for ds in datasets:
  248. ds_data = [r for r in all_data if r.get('dataset') == ds]
  249. if not ds_data:
  250. continue
  252. # Get sample-level info for this dataset
  253. ds_info = ds_data[0] # All records share same dataset structure
  254. n_total = ds_info.get('n_ctrl', 0) + ds_info.get('n_pd', 0)
  256. # For simplicity, shuffle Δr values across pairs within dataset
  257. # This preserves within-dataset structure
  258. for r in ds_data:
  259. perm_records.append(r.copy())
  261. # Actually: proper permutation shuffles sample labels, recomputes correlations
  262. # But that's very expensive. Instead, we shuffle Δr assignments across categories
  263. # while preserving dataset structure.
  264. #
  265. # Simpler valid approach: randomly reassign category labels
  266. np.random.shuffle(perm_records)
  268. # Recompute selectivity on shuffled data
  269. perm_sel = compute_selectivity(perm_records, datasets)
  270. if perm_sel is not None:
  271. perm_sels.append(perm_sel['selectivity'])
  273. perm_sels = np.array(perm_sels)
  274. p_value = np.mean(np.abs(perm_sels) >= np.abs(obs_sel))
  276. return {
  277. 'observed_selectivity': obs_sel,
  278. 'perm_mean': np.mean(perm_sels),
  279. 'perm_std': np.std(perm_sels),
  280. 'p_value': p_value,
  281. 'n_perm': len(perm_sels),
  282. }
  285. # ---------------------------------------------------------------------------
  286. # Main Pipeline
  287. # ---------------------------------------------------------------------------
  288. def main():
  289. print("=" * 80)
  290. print("ALDH1A1-PD META-ANALYSIS: Script 03 — Cell Type Enrichment & Selectivity")
  291. print("=" * 80)
  293. target_pairs = list(combinations(TARGET_GENES, 2))
  295. deconv_results = []
  296. all_corr_records = []
  298. for gse_id in DECONV_DATASETS:
  299. info = DATASETS[gse_id]
  300. print(f"\n{'=' * 60}")
  301. print(f" Processing {gse_id}")
  302. print(f"{'=' * 60}")
  304. try:
  305. gse = download_series_matrix(gse_id)
  307. if gse_id == 'GSE8397':
  308. expr_df, groups = process_GSE8397(gse)
  309. elif gse_id == 'GSE49036':
  310. expr_df, groups = process_GSE49036(gse)
  311. else:
  312. expr_df, groups = process_standard_dataset(gse, gse_id)
  314. # Log2 transform if needed
  315. if expr_df.max().max() > 100:
  316. expr_df = np.log2(expr_df.clip(lower=1))
  318. ctrl_samples = [s for s, g in groups.items() if g == 'control' and s in expr_df.columns]
  319. pd_samples = [s for s, g in groups.items() if g == 'PD' and s in expr_df.columns]
  321. n_ctrl = len(ctrl_samples)
  322. n_pd = len(pd_samples)
  323. print(f" Samples: {n_ctrl} ctrl + {n_pd} PD")
  325. # --- Enrichment scoring ---
  326. scores, markers_used = compute_enrichment_scores(expr_df, CELL_TYPE_SIGNATURES)
  328. for ct, (avail, total) in markers_used.items():
  329. print(f" {ct}: {avail}/{total} markers")
  331. # --- Cell type differences ---
  332. for ct in CELL_TYPE_SIGNATURES:
  333. if ct not in scores.columns:
  334. continue
  335. ctrl_scores = scores.loc[ctrl_samples, ct].dropna()
  336. pd_scores = scores.loc[pd_samples, ct].dropna()
  338. if len(ctrl_scores) < 3 or len(pd_scores) < 3:
  339. continue
  341. t_val, p_val = stats.ttest_ind(ctrl_scores, pd_scores, equal_var=False)
  342. pooled = np.sqrt((ctrl_scores.var() * (len(ctrl_scores)-1) +
  343. pd_scores.var() * (len(pd_scores)-1)) /
  344. (len(ctrl_scores) + len(pd_scores) - 2))
  345. d = (ctrl_scores.mean() - pd_scores.mean()) / pooled if pooled > 0 else 0
  347. deconv_results.append({
  348. 'dataset': gse_id,
  349. 'cell_type': ct,
  350. 'ctrl_mean': ctrl_scores.mean(),
  351. 'pd_mean': pd_scores.mean(),
  352. 'p_value': p_val,
  353. 'cohens_d': d,
  354. 'n_ctrl': len(ctrl_scores),
  355. 'n_pd': len(pd_scores),
  356. 'significant': p_val < 0.05,
  357. })
  359. # --- Raw and adjusted correlations ---
  360. da_vuln_scores = scores['DA_Vulnerable_SOX6'] if 'DA_Vulnerable_SOX6' in scores.columns else None
  362. for g1, g2 in target_pairs:
  363. if g1 not in expr_df.index or g2 not in expr_df.index:
  364. continue
  366. pair_key = (g1, g2)
  367. pair_key_rev = (g2, g1)
  368. cat = GENE_PAIR_CATEGORIES.get(pair_key, GENE_PAIR_CATEGORIES.get(pair_key_rev, 'Other'))
  370. # Raw correlations
  371. x_ctrl = expr_df.loc[g1, ctrl_samples].astype(float).values
  372. y_ctrl = expr_df.loc[g2, ctrl_samples].astype(float).values
  373. x_pd = expr_df.loc[g1, pd_samples].astype(float).values
  374. y_pd = expr_df.loc[g2, pd_samples].astype(float).values
  376. r_ctrl, _ = stats.pearsonr(x_ctrl[~np.isnan(x_ctrl) & ~np.isnan(y_ctrl)],
  377. y_ctrl[~np.isnan(x_ctrl) & ~np.isnan(y_ctrl)])
  378. r_pd, _ = stats.pearsonr(x_pd[~np.isnan(x_pd) & ~np.isnan(y_pd)],
  379. y_pd[~np.isnan(x_pd) & ~np.isnan(y_pd)])
  380. raw_dr = r_pd - r_ctrl
  382. # Adjusted correlation (partial, controlling for DA_Vulnerable)
  383. adj_dr = np.nan
  384. if da_vuln_scores is not None:
  385. z_ctrl = da_vuln_scores[ctrl_samples].values
  386. z_pd = da_vuln_scores[pd_samples].values
  388. r_adj_ctrl = partial_corr(x_ctrl, y_ctrl, z_ctrl)
  389. r_adj_pd = partial_corr(x_pd, y_pd, z_pd)
  391. if not np.isnan(r_adj_ctrl) and not np.isnan(r_adj_pd):
  392. adj_dr = r_adj_pd - r_adj_ctrl
  394. record = {
  395. 'dataset': gse_id,
  396. 'pair_name': f"{g1}-{g2}",
  397. 'gene1': g1,
  398. 'gene2': g2,
  399. 'category': cat,
  400. 'r_ctrl': r_ctrl,
  401. 'r_pd': r_pd,
  402. 'delta_r': raw_dr,
  403. 'adj_delta_r': adj_dr,
  404. 'n_ctrl': n_ctrl,
  405. 'n_pd': n_pd,
  406. }
  407. all_corr_records.append(record)
  409. except Exception as e:
  410. print(f" ERROR: {e}")
  411. import traceback
  412. traceback.print_exc()
  413. continue
  415. # --- Selectivity Analysis ---
  416. print("\n" + "=" * 80)
  417. print("SELECTIVITY ANALYSIS")
  418. print("=" * 80)
  420. # Raw selectivity (6 datasets)
  421. print("\n--- Raw Selectivity (6 deconv datasets) ---")
  422. raw_sel = compute_selectivity(all_corr_records)
  423. if raw_sel:
  424. for k, v in raw_sel.items():
  425. print(f" {k}: {v}")
  427. # Adjusted selectivity
  428. print("\n--- Adjusted Selectivity (6 deconv datasets) ---")
  429. adj_records = []
  430. for r in all_corr_records:
  431. adj_r = r.copy()
  432. adj_r['delta_r'] = r.get('adj_delta_r', r['delta_r'])
  433. adj_records.append(adj_r)
  435. adj_sel = compute_selectivity(adj_records)
  436. if adj_sel:
  437. for k, v in adj_sel.items():
  438. print(f" {k}: {v}")
  440. # 4-validated dataset selectivity
  441. validated = ['GSE8397', 'GSE20163', 'GSE20164', 'GSE49036']
  442. print("\n--- 4-Validated Dataset Selectivity ---")
  443. val_sel = compute_selectivity(all_corr_records, validated)
  444. if val_sel:
  445. for k, v in val_sel.items():
  446. print(f" {k}: {v}")
  448. # Permutation test
  449. print("\n--- Permutation Test ---")
  450. perm_result = permutation_test(all_corr_records, n_perm=5000)
  451. if perm_result:
  452. for k, v in perm_result.items():
  453. print(f" {k}: {v}")
  455. # --- Save Results ---
  456. deconv_df = pd.DataFrame(deconv_results)
  457. deconv_df.to_csv(os.path.join(RESULTS_DIR, 'deconvolution_results.csv'), index=False)
  459. corr_df = pd.DataFrame(all_corr_records)
  460. corr_df.to_csv(os.path.join(RESULTS_DIR, 'raw_vs_adjusted_correlations.csv'), index=False)
  462. sel_data = []
  463. if raw_sel:
  464. sel_data.append({'analysis': '6-dataset raw', **raw_sel})
  465. if adj_sel:
  466. sel_data.append({'analysis': '6-dataset adjusted', **adj_sel})
  467. if val_sel:
  468. sel_data.append({'analysis': '4-validated raw', **val_sel})
  469. if perm_result:
  470. sel_data.append({'analysis': 'permutation', **perm_result})
  472. pd.DataFrame(sel_data).to_csv(os.path.join(RESULTS_DIR, 'selectivity_analysis.csv'), index=False)
  474. print(f"\nResults saved to {RESULTS_DIR}/")
  477. if __name__ == '__main__':
  478. main()

03_cell_type_enrichment.py at commit f9b5f19, under MIT · at the source

Overview

Authors: Drake H Harbert1
  1. Inner Architecture LLC, Canton, OH, United States
Journal: Frontiers in aging neuroscience, volume 18, article 1806505
Dates: received 7 February 2026; accepted 16 April 2026; published online 19 May 2026
Type: Systematic review · Language: English
License: CC BY
Identifiers: DOI 10.3389/fnagi.2026.1806505 · PMID 42239820 · PMCID PMC13226206 · OpenAlex W7161641621
Open access: gold, a free copy (OpenAlex)
Status: code verified
Categories: human (organism), Parkinson's (population), cellular / molecular (subfield)
Methods: Spectral & time-frequency, Statistics, Preprocessing, Connectivity
Keywords: ALDH1A1, alpha-synuclein, cell type enrichment, dopamine, gene expression, meta-analysis, neurodegeneration, Parkinson’s disease
Topic: Parkinson's Disease Mechanisms and Treatments (Neurology, Medicine), according to OpenAlex
Citations: not cited yet (Europe PMC); 29 references in the paper

Abstract

Background: Parkinson’s disease (PD) involves progressive dopaminergic neuron loss in the substantia nigra (SN). Aldehyde dehydrogenase 1A1 (ALDH1A1), the rate-limiting enzyme in retinoic acid biosynthesis, is enriched in vulnerable dopaminergic neuron subpopulations and is consistently downregulated in PD. However, the relationship between ALDH1A1 expression and broader dopaminergic pathway gene co-expression has not been systematically characterized across multiple independent datasets.

Methods: Gene expression correlations were analyzed across seven independent human SN microarray datasets (n = 156; 70 controls, 86 PD) from the Gene Expression Omnibus. Simple arithmetic means across datasets are reported as the primary summary statistic; random-effects meta-analysis with DerSimonian-Laird estimation was applied to Fisher’s z-transformed correlation coefficients to generate pooled estimates with heterogeneity statistics. Marker gene-based enrichment scoring using published cell type markers from single-nucleus RNA-seq profiling of human substantia nigra—with all target genes excluded from signatures—was performed across six analyzable datasets. Selectivity of ALDH1A1 correlation attenuation was assessed using permutation testing (n = 5,000) as the primary statistical test, with parametric tests reported as supplementary.

Results: In controls, ALDH1A1 showed strong co-expression with dopaminergic genes (mean r = 0.92–0.93 for TH, DDC, and SLC18A2). In PD, these correlations were attenuated (mean Δr = −0.336 for ALDH1A1-dopamine pairs). Dopamine-dopamine correlations showed less attenuation (mean Δr = −0.143). Marker gene-based enrichment scoring confirmed significant depletion of ALDH1A1-positive vulnerable dopaminergic neurons in 4 of 6 datasets. After adjusting for estimated cell type enrichment, the selectivity of ALDH1A1 attenuation was preserved (adjusted selectivity: −0.210, increased from raw selectivity of −0.190; raw permutation p = 0.0052).

Conclusion: ALDH1A1 co-expression with dopaminergic pathway genes is attenuated in PD substantia nigra across all seven datasets examined. This attenuation is selective for ALDH1A1-containing pairs, and this selectivity persists after adjusting for cell type enrichment changes. While correlational, these findings are consistent with a role for retinoic acid pathway disruption in PD pathophysiology and warrant mechanistic investigation.

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 12 matches between paragraphs and lines of code.

nwharbert8-ui/ALDH1A1-PD-meta-analysis-Repo

License: MIT
State: the link answers, verified on 28 September 2026
Evidence: files inventoried
Commit: f9b5f19231ba7fb507e61698f5505b8fba9f3f08, 7 February 2026
Languages: Python (6)
Size: 9 files, 6 scripts
Software Heritage: not archived
Found in: “Data availability statement”
Holds: README, license file, environment (requirements.txt)
Not found: CITATION.cff, tests, continuous integration, documentation
Tools: NumPy (5 files), pandas (5 files), SciPy (5 files), Matplotlib (1 file)
Availability: 1 check, the latest on 28 September 2026: the link answers
  • 28 September 2026: the link answers
8 files
At the source: github.com/nwharbert8-ui/ALDH1A1-PD-meta-analysis-Repo

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;
  • 12 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 statement

Publicly available datasets were analyzed in this study. The GEO datasets can be found at the following URLs: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE7621; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE8397; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE20163; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE20164; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE20292; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE20333; and https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE49036. The complete analysis code for this meta-analysis is publicly available at https://github.com/nwharbert8-ui/ALDH1A1-PD-meta-analysis-Repo.

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 1, 28 September 2026: the first record

Recorded: type, language, journal, volume, pages, dates, 1 author, 8 keywords, 29 references.

Cite

This paper

Harbert, D. H. (2026). ALDH1A1-dopaminergic gene co-expression in human substantia nigra: meta-analysis of disease-associated correlation changes across seven independent Parkinson's disease datasets. Frontiers in aging neuroscience, 18, 1806505. https://doi.org/10.3389/fnagi.2026.1806505

BibTeX

@article{harbert2026aldh1a1,
author = {Harbert, Drake H},
title = {{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},
year = {2026},
month = may,
volume = {18},
pages = {1806505},
publisher = {Frontiers Media SA},
issn = {1663-4365},
doi = {10.3389/fnagi.2026.1806505},
url = {https://doi.org/10.3389/fnagi.2026.1806505},
pmid = {42239820},
pmcid = {PMC13226206}
}

RIS

TY - JOUR
AU - Harbert, Drake H
TI - ALDH1A1-dopaminergic gene co-expression in human substantia nigra: meta-analysis of disease-associated correlation changes across seven independent Parkinson's disease datasets
T2 - Frontiers in aging neuroscience
J2 - Front Aging Neurosci
PY - 2026
DA - 2026/05/19
VL - 18
SP - 1806505
SN - 1663-4365
PB - Frontiers Media SA
DO - 10.3389/fnagi.2026.1806505
UR - https://doi.org/10.3389/fnagi.2026.1806505
LA - en
ER -

CSL-JSON

{
"id": "10.3389/fnagi.2026.1806505",
"type": "article-journal",
"title": "ALDH1A1-dopaminergic gene co-expression in human substantia nigra: meta-analysis of disease-associated correlation changes across seven independent Parkinson's disease datasets",
"container-title": "Frontiers in aging neuroscience",
"author": [
{
"family": "Harbert",
"given": "Drake H"
}
],
"container-title-short": "Front Aging Neurosci",
"volume": "18",
"page": "1806505",
"DOI": "10.3389/fnagi.2026.1806505",
"PMID": "42239820",
"PMCID": "PMC13226206",
"ISSN": "1663-4365",
"publisher": "Frontiers Media SA",
"URL": "https://doi.org/10.3389/fnagi.2026.1806505",
"language": "en",
"issued": {
"date-parts": [
[
2026,
5,
19
]
]
}
}

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:10.1371/journal.pcbi.1014323 [code]
A prototype-augmented graph representation learning framework for identifying brain disorder-associated genes and facilitating drug repurposing.
Journal: PLoS computational biology
In common: pandas, Matplotlib, NumPy, NCBI GEO GSE7621, Parkinson's, cellular / molecular, 1 reference
[2] doi:10.1073/pnas.2613593123 [code]
Calbindin stratifies midbrain dopaminergic neurons governing distinct aspects of locomotion.
Journal: Proceedings of the National Academy of Sciences of the United States of America
In common: Parkinson's, 5 references
[3] doi:10.1038/s41467-026-75194-3 [code]
Leucine-rich repeat kinase 2 impairs the release sites of Parkinson's disease vulnerable dopamine axons.
Journal: Nature communications
In common: Parkinson's, cellular / molecular, 4 references
[4] doi:10.3389/fnagi.2026.1847611 [code]
APOE ε4-associated hippocampal atrophy trajectories across the Alzheimer's disease continuum: a systematic review, meta-analysis, and longitudinal validation.
Journal: Frontiers in aging neuroscience
In common: pandas, SciPy, Matplotlib, 1 other tool, 3 references
[5] doi:10.1016/j.stemcr.2026.102930 [code]
ZFHX4 is necessary for dopaminergic neuron differentiation and controls cell cycle by regulating LIN28A.
Journal: Stem cell reports
In common: pandas, SciPy, Matplotlib, 1 other tool, Parkinson's, cellular / molecular, 2 references
[6] doi:10.1002/cns.71075
Neuroprotective Role of E3 Ubiquitin Ligase TRIM2 in Parkinson's Disease: Attenuation of Oxidative Stress and Apoptosis via Promoting ELAVL1 Ubiquitination.
Journal: CNS neuroscience & therapeutics
In common: NCBI GEO GSE7621, Parkinson's, cellular / molecular
[7] doi:10.1038/s41593-026-02316-x [code]
Single-cell multi-omic atlas and morphogen screening informs midbrain and hindbrain organoid engineering.
Journal: Nature neuroscience
In common: pandas, SciPy, Matplotlib, 1 other tool, 2 references
[8] doi:10.1038/s41380-026-03667-4
Engineering functional ventral midbrain dopaminergic neurons in human organoids through WNT modulation and bioreactor culture.
Journal: Molecular psychiatry
In common: Parkinson's, cellular / molecular, 3 references
[9] doi:10.1172/jci190954
Modulation of WNT and FGF18 enhances yield and subtype identity of hPSC-derived midbrain dopamine neurons.
Journal: The Journal of clinical investigation
In common: Parkinson's, 3 references
[10] doi:10.1016/j.ebiom.2026.106293 [code]
Dynamic neural states underpin motor symptom severity in Parkinson's disease: a longitudinal analysis of chronic cortico-subthalamic nucleus recordings.
Journal: EBioMedicine
In common: pandas, SciPy, Matplotlib, 1 other tool, Parkinson's, 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.