git-nome — 41× Whole-Genome Sequencing, Multi-Trait Bioinformatic Pipeline & Genomic Architecture

An open-science, reproducible clinical and population-genomic reference implementation by Mario Marcolongo (ENA PRJEB109744 / BioSample SAMEA121950568).

---

1. Overview & Architectural Vision

Conventional clinical genomics operates in proprietary black boxes: consumer DNA panels evaluate less than 0.02% of the genome, while standard clinical exome sequencing ignores non-coding regulatory networks, structural variants, and complex polygenic interactions.

This repository represents an end-to-end, reproducible open-science genomic pipeline. Instead of relying on commercial summaries, this infrastructure processes raw 41× Whole-Genome Sequencing (WGS) reads aligned to the human reference genome (GRCh38/hg38), executing multi-layered quality control, high-precision ancestry standardizations, pharmacogenomic profiling, and variant-to-ontology mapping across both cloud (Terra.bio) and local Apple Silicon (macOS arm64) environments.

The Atypical / Neurodivergent Engineering Edge

This pipeline was engineered by a neurodivergent mind driven by bottom-up verification, uncompromising empirical rigor, and deep systems pattern recognition. Where standard clinical workflows settle for pre-packaged commercial reporting, this architecture cross-examines raw .vcf.gz variants directly against primary biomedical literature, gnomAD constraint metrics (pLI/LOEUF), HGDP+1kGP reference panels, and structured disease ontologies (HPO/MONDO)—embodying an uncompromising standard of open-science verification.

---

2. Comprehensive Multi-Layer Genome Analysis Pipeline

Beyond standard Polygenic Risk Scoring, the git-nome pipeline implements modular, highly specialized bioinformatic extraction and annotation modules:

graph TD
    A[Raw 41× WGS FASTQ/BAM<br/>ENA PRJEB109744] --> B[Terra.bio Upstream Alignment<br/>GRCh38 / hg38 Coordinates]
    B --> C[Plink2 High-Performance Conversion<br/>PGEN / PVAR / PSAM]
    C --> D[PGS Catalog Calculator<br/>Nextflow pgsc_calc]
    C --> E[Custom Python/Shell Extraction Modules]
    
    D --> F[HGDP + 1kGP Reference Panel<br/>PCA Projection & Z-Score Standardization]
    
    E --> G[ACMG Secondary Findings Screening<br/>acmg_secondary_findings.py]
    E --> H[Mitochondrial Heteroplasmy & HaploGrep3<br/>extract_mt_clinical_enhanced.py]
    E --> I[Y-Chromosome Patrilineal Profiling<br/>analyze_y_haplogroup.py]
    E --> J[Dark Matter & Novel Variant Mapping<br/>extract_dark_matter.py / HPO & MONDO]
    E --> K[Pharmacogenomics & Endocrine Response<br/>extract_lifestyle_pharma.py / CYP & VDR]
    E --> L[Pleiotropic & Cognitive Network Loci<br/>extract_precision_math.py / Targeted Panels]

Loading

Module Catalog & Methodology

1. Ancestry-Adjusted Polygenic Risk Scoring (pgsc_calc + HGDP+1kGP)

• Core Engine: Executes the standard Nextflow pgsc_calc pipeline against Plink2 binary genotypes (.pgen).
• Unified Comparison Catalog: To eliminate population stratification bias, scores are standardized via Principal Component Analysis (PCA) projection onto the Broad Institute HGDP + 1kGP (Human Genome Diversity Project + 1000 Genomes Project) reference panel (~3,400+ high-coverage genomes across 54 global populations).
• Output Metrics: Computes ancestry-standardized $Z$-scores (Z_MostSimilarPop) and exact match-rate audits (min_overlap: 0.75).

2. Actionable Clinical Screening (acmg_secondary_findings.py)

• Automated interrogation of the American College of Medical Genetics (ACMG SF v3.2) actionable gene list (BRCA1, BRCA2, LDLR, MYBPC3, RYR1, KCNQ1, MLH1, MSH2, etc.).
• Evaluates functional impact, ClinVar pathogenic assertions (clinvar.vcf.gz), and zygosity to flag high-impact medically actionable findings.

3. Mitochondrial Heteroplasmy & Lineage Analysis (extract_mt_clinical_enhanced.py, analyze_mt_haplogroup.py)

• High-depth variant extraction across the circular mitochondrial genome (MT / rCRS).
• Integrates with HaploGrep3 (phylotree-rcrs) for maternal haplogroup classification and assesses clinical heteroplasmy thresholds against MITOMAP and ClinVar mitochondrial datasets.

4. Patrilineal Y-Chromosome Classification (analyze_y_haplogroup.py)

• Systematic interrogation of non-recombining Y-chromosome (NRY) single-nucleotide polymorphisms against established ISOGG and Y-Full phylogenetic markers.

5. Genomic Dark Matter & Novel Structural Variants (extract_dark_matter.py, scan_syndromic_sv.py)

• Identifies unannotated, rare, or non-coding structural variants, large copy-number variations (CNVs), and regulatory disruption motifs.
• Cross-references unmapped variants against Human Phenotype Ontology (HPO) and MONDO Disease Ontology to surface potential candidate pleiotropic loci.

6. Pharmacogenomics (PGx) & Endocrine Pathways (extract_lifestyle_pharma.py, extract_vitamin_d.py)

• Profiles Phase I/II drug metabolism enzymes and transporters (CYP2D6, CYP2C19, CYP3A4, CYP2C9, SLCO1B1, DPYD, VKORC1).
• Evaluates metabolic, dietary, and endocrine response pathways, including Vitamin D receptor (VDR), binding protein (GC), and hydroxylation kinetics.

7. Precision Cognitive & Neurodevelopmental Networks (extract_precision_math.py, extract_cognitive_variants.py, extract_targeted_panels.py)

• Deep-dive screening of targeted gene panels across neurodevelopmental (ASD), connective tissue (CTD), and complex pleiotropic cognitive networks.

8. Evolutionary Constraint & Conservation (enrich_constraint.py)

• Annotates variants with gnomAD loss-of-function intolerance (pLI), loss-of-function observed/expected upper bound fraction (LOEUF), and mammalian evolutionary conservation scores (phyloP, phastCons).

9. Deep Evolutionary Adaptations & Environmental Resilience (extract_evolutionary_adaptations.py)

• Automated bioinformatic interrogation of human evolutionary adaptation suites across high-coverage whole genome data.
• Hypoxia Pathways: Interrogates Tibetan (EPAS1, EGLN1, HIF1A) and Andean (DST right ventricle adaptation, MGAM starch metabolism, CD83/IL-36R pathogen bottleneck markers) suites.
• Submergence & Vasoconstriction: Interrogates the Bajau diving reflex (BDKRB2 peripheral vasoconstriction, PDE10A spleen volume regulation).
• Arsenic Methylation & Environmental Tolerance: Screens Chromosome 10q24.32 for the protective AS3MT Met287Thr C-T-T-A tolerance haplotype and FTCD.
• Connective Tissue & Extracellular Matrix: Interrogates sports injury, tendon remodeling, and mitochondrial defense markers (COL1A1 Sp1, MMP3, GDF5, CKMM, SOD2).
• Viral Resistance: Screens CCR5-delta32, HCP5 proxy (HLA-B*57:01 elite controller status), and CCR2-V64I.

---

3. Data Governance & Scientific Integrity: Why Methodology is Public and Behavioral Scores are Excluded

A core philosophy of this open-science repository is uncompromising scientific precision and ethical data governance. While the architecture, bioinformatic scripts, parameter schemas, and Nextflow workflows are published openly (AGPL-3.0), individual computed personal reports—particularly Polygenic Risk Scores for psychiatric and cognitive traits (Schizophrenia, Bipolar Disorder, Depression, ADHD, IQ/Fluid Intelligence)—are intentionally excluded from the public git tracking tree via .gitignore.

This decision rests on three distinct pillars of biomedical rigor and data stewardship:

1. Scientific Validity over False Determinism: Current research-stage polygenic risk scores for behavioral traits explain only a fraction of phenotypic variance and are heavily influenced by linkage disequilibrium and cohort design. Publishing rendered chart images or static tables of probabilistic personal scores risks them being misread by lay evaluators, automated background screening tools, or AI crawlers as deterministic clinical diagnoses or fixed limits.
2. Familial Privacy Stewardship: Genomic variants are shared biologically with parents, siblings, and offspring (50% inheritance). Actionable clinical findings and psychiatric risk distributions inherently carry probabilistic implications for biological relatives who did not consent to public disclosure.
3. Reproducibility Verification: Proving that an engineering pipeline works does not require exposing private individual traits. This pipeline is validated cleanly against standardized open reference samples (such as the 1000 Genomes NA12878/HG001 benchmark genomes) and low-stakes physiological reference traits (such as Standing Height and Chronotype), delivering 100% of the technical demonstration value while setting a high ethical standard for open biomedical data.

---

4. Repository Structure & Quick-Start Guide

Key Directories & Files

• GENETIC_ANALYSIS_GUIDE.md — Detailed technical manual for biotechnologists running the pgsc_calc pipeline.
• predict_personal.py & visualize_pgs.py — Core Python engines for trait prediction and normal distribution rendering against HGDP+1kGP.
• acmg_secondary_findings.py — Automated ACMG SF v3.2 clinical gene extraction script.
• extract_*.py & analyze_*.py — Specialized domain extraction modules (MT, Y, Dark Matter, PGx, Precision Math).
• run_*.sh & setup_local_vep.sh — Execution shell scripts for batch processing and local Apple Silicon environment provisioning.
• pgsc_params_*.json — Config schemas defining Nextflow parameters (target_build, min_overlap, ancestry).

Execution Instructions

1. Provision Environment:

   ./setup_local_vep.sh

2. Run Batch Polygenic Scoring (pgsc_calc):

   ./run_batch_analysis.sh

3. Execute Actionable Clinical & Domain Extractions:

   python acmg_secondary_findings.py
   python extract_dark_matter.py
   python extract_lifestyle_pharma.py

4. Generate Visualization Charts & Summary Reports:

   python visualize_pgs.py
   python generate_clinical_dashboard.py

---

5. Public Accessions & Open Science License

• European Nucleotide Archive (ENA) Study Accession: PRJEB109744
• BioSample Accession: SAMEA121950568
• Code & Methodology License: Dedicated under AGPL-3.0 / CC BY-SA 4.0.
• Raw WGS Data License: Dedicated to the public domain under CC0 1.0 Universal (PRJEB109744).

Note on Large Files & GitHub Limits: Raw alignment .bam files (~30GB), high-coverage .vcf datasets (1.4GB), and HGDP+1kGP reference panels (7.4GB .tar.zst) are hosted on ENA (PRJEB109744 / SAMEA121950568) to comply with GitHub's 100MB file limit. All pipeline code, extraction modules, configuration schemas, and methodology documentation are tracked directly within this repository.