Testing whether the ratio of nuclear-to-mtDNA OXPHOS gene expression rises with age across human tissues, whether that imbalance predicts suppression of a regeneration gene signature, whether one "chokepoint" gene drives it, and whether that gene is druggable.
| Phase | Question | Headline result |
|---|---|---|
| 0 | Environment setup | Done |
| 1 | Does imbalance rise with age? | Tissue-specific: up in muscle/skin, down in brain/heart/liver |
| 2 | Does imbalance suppress regeneration genes? | Yes in brain/liver/heart (partial ρ up to −0.43), no effect in muscle, reversed in skin |
| 3 | Is there one chokepoint gene? | No single universal gene; NAMPT is the only candidate ranking top-15/135 in all 5 tissues |
| 4 | Is NAMPT druggable with NAD⁺ precursors? | Nicotinamide (−6.70 kcal/mol) and NMN (−7.53) dock favorably and are mechanistically valid; NR (−8.46) is likely a structural-mimicry false positive |
| 5 | Cross-species check (killifish) | Blocked: both public datasets checked lack mtDNA gene quantification |
Full write-up with dual-tier (plain-language + technical) explanations, tool rationale, and future directions: see docs/PROJECT_GUIDE.md.
mito-aging-project/
├── scripts/ # all analysis scripts, in run order (01_... through 23_...)
├── results/ # output tables (scores, docking results, merged datasets)
├── figures/ # plots
├── data/ # raw/intermediate data (not pushed, see .gitignore; scripts regenerate it)
└── docs/ # full project guide
Environment: Conda (mito-aging env, Python 3.10 + R 4.3/4.6); see the scripts for the package list. Scripts are numbered in intended run order, and each NN_*.py / .R file corresponds to one step described in docs/PROJECT_GUIDE.md.
Python (pandas, scipy, scikit-learn, shap, statsmodels, rdkit, meeko), R/Bioconductor (recount3, recount, DESeq2, biomaRt, tidyverse), AutoDock Vina, Open Babel, MitoCarta 3.0, GTEx (via recount3), GEO.