Distinct Long Non-coding RNA Regulatory Architectures Encode Lineage Stability and Epithelial Plasticity
收藏资源简介：
# lncRNA Regulatory Architecture Study **Distinct Long Non-coding RNA Regulatory Architectures Encode Lineage Stability and Epithelial Plasticity** **Author:** Neetu Singh, PhD **Affiliation:** Associate Professor, Molecular Biology Unit, Center for Advance Research, King George's Medical University, Lucknow-226 003, Uttar Pradesh, India **Contact:** neetusingh@kgmcindia.edu **ORCID:** [add your ORCID here] > **Manuscript DOI:** [add after acceptance] > **Processed Data (Zenodo):** [https://doi.org/10.5281/zenodo.XXXXXXX](https://doi.org/10.5281/zenodo.XXXXXXX) > **Code Archive (Zenodo):** [https://doi.org/10.5281/zenodo.YYYYYYY](https://doi.org/10.5281/zenodo.YYYYYYY) --- ## Overview This repository contains the complete analysis pipeline for an integrative study of ~70,500 annotated lncRNA transcripts across hematopoietic, brain, endometrial, and breast tissues profiled on the Affymetrix Human Transcriptome Array 2.0 (HTA-2.0, GPL17586). **Core findings:**- Hematopoietic, brain, and endometrial samples form compact lineage-resolved PCA clusters (stable regulatory programs)- Breast samples show broad dispersion and high loading entropy (distributed, plastic regulatory architecture)- Only 33 lncRNAs are shared between brain and endometrium; no overlap across all four lineages- Cohort-imbalance controls confirmed hematopoietic PC1 dominance is biological, not artefactual (permutation p < 0.001) --- ## Repository Structure ```lncrna-lineage-architecture/├── scripts/│ ├── 01_preprocessing.py # Merge GEO CSVs → KNN imputation → z-score│ ├── 02_pca_analysis.py # PCA + lineage lncRNA selection + cohort controls│ ├── 03_entropy_upset_heatmap.py # Shannon entropy (bootstrapped CI) + UpSet + heatmap│ ├── 04_umap_batch_assessment.py # UMAP batch-effect visualisation (Supp. Fig. S14)│ ├── download_geo_data.sh # Auto-download raw data from NCBI GEO│ └── requirements.txt # Python package dependencies│├── data/│ ├── raw/ # Raw GEO expression CSVs (NOT tracked — download from GEO)│ └── processed/ # Processed matrices (available on Zenodo — link above)│├── results/│ ├── pca/ # PCA scores, loadings, lineage probe-set lists, stat tests│ ├── entropy/ # Entropy scores with bootstrapped CIs│ ├── umap/ # UMAP coordinates│ └── figures/ # All manuscript figures (600 dpi PNG + vector PDF)│├── supplementary/│ └── Supplementary_Tables_S1_S4.xlsx # Lineage-associated lncRNA lists│├── .gitignore├── LICENSE└── README.md``` --- ## Source Data (NCBI GEO) All raw expression data are publicly available. Replace `GSE######` with real accessions: | GEO Accession | Tissue | Condition | Platform | n Samples ||---------------|-------------|-----------|----------|-----------|| GSE###### | Hematopoietic | MDS | GPL17586 | ~207 || GSE###### | Breast | Tumour | GPL17586 | ~75 || GSE###### | Breast | Normal | GPL17586 | ~13 || GSE###### | Brain | Tumour | GPL17586 | ~7 || GSE###### | Brain | Normal | GPL17586 | ~2 || GSE###### | Endometrium | Benign | GPL17586 | ~8 | --- ## Processed Data (Zenodo) The four processed files required to reproduce downstream analyses are deposited at: **https://doi.org/10.5281/zenodo.XXXXXXX** | File | Description ||------|-------------|| `expression_matrix_imputed.csv` | Merged HTA-2.0 matrix after KNN imputation (k=5) || `expression_matrix_zscore.csv` | Z-scored matrix — input to all PCA analyses || `sample_metadata.csv` | Sample annotations with GEO accessions || `probe_manifest.csv` | Retained probe set IDs (~70,500) | --- ## Reproducing the Analysis ### Setup ```bashgit clone https://github.com/YourUsername/lncrna-lineage-architecture.gitcd lncrna-lineage-architecture python -m venv venvsource venv/bin/activate # Windows: venv\Scripts\activatepip install -r scripts/requirements.txt``` ### Option A — Use processed data from Zenodo (fastest) ```bashmkdir -p data/processed# Download the 4 CSV files from https://doi.org/10.5281/zenodo.XXXXXXX# Place them in data/processed/``` ### Option B — Reproduce from raw GEO data ```bash# 1. Edit download_geo_data.sh — replace all GSE###### with real accessions# 2. Run the downloaderbash scripts/download_geo_data.sh # 3. Run preprocessingpython scripts/01_preprocessing.py``` ### Run the full pipeline ```bashpython scripts/02_pca_analysis.py # PCA + cohort controls + Levene's testpython scripts/03_entropy_upset_heatmap.py # Entropy + UpSet + heatmappython scripts/04_umap_batch_assessment.py # UMAP batch assessment``` All figures are saved to `results/figures/` as both PNG (600 dpi) and PDF. --- ## Analysis Parameters | Parameter | Value | Reference ||-----------|-------|-----------|| KNN k | 5 | Supp. Method S6 || Loading threshold | Top 2% (~1,410 probesets) | Supp. Method S1 || Bootstrap iterations | 1,000 | Supp. Method S4 || Permutation iterations | 1,000 | Supp. Method S4 || Down-sampling iterations | 100 | Supp. Method S4 || UMAP n_neighbors | 15 | Supp. Method S5 || UMAP min_dist | 0.1 | Supp. Method S5 || Random seed | 42 | Throughout | --- ## Citation ```bibtex@article{Singh2025lncRNA, author = {Singh, Neetu}, title = {Distinct Long Non-coding RNA Regulatory Architectures Encode Lineage Stability and Epithelial Plasticity}, journal = {[Journal]}, year = {2025}, doi = {[pending]}}``` --- ## License Code: [MIT License](LICENSE) Processed data (Zenodo): [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)
