Biomarker Data Science Platform

End-to-End Clinical Trial Biomarker Pipelines across Therapeutic Areas

Reproducible, modular biomarker analysis reports spanning four therapeutic areas. Each report applies an identical analytical pipeline to simulated Phase II trial data, with disease-specific biological framing and clinical interpretation.

Therapeutic Area Reports

🔵 Immunology — FcRn Inhibitor

Disease · Generalised Myasthenia Gravis (gMG)
Mechanism · FcRn-mediated IgG catabolism (VYVGART-like)
Primary biomarker · IgG total NPX
N · 120 patients

🔴 Oncology — Checkpoint Inhibitor

Disease · Non-Small Cell Lung Cancer (NSCLC)
Mechanism · PD-1 blockade / T-cell reinvigoration (nivolumab-like)
Primary biomarker · Tumour Mutational Burden (TMB)
N · 150 patients

🟢 Cardiovascular — PCSK9 Inhibitor

Disease · Heterozygous Familial Hypercholesterolaemia (HeFH)
Mechanism · PCSK9 inhibition / LDL-R upregulation (evolocumab-like)
Primary biomarker · LDL Cholesterol (LDL-C)
N · 200 patients

🟣 Neurology — Anti-Amyloid

Disease · Early Alzheimer’s Disease (MCI / mild dementia)
Mechanism · Anti-Aβ protofibril mAb (lecanemab-like)
Primary biomarker · Plasma p-tau181
N · 160 patients

Pipeline Architecture

All reports share a common three-stage analytical framework executed from a single parameterised source file (_analysis_core.qmd):

Stage	Methods
Multi-Omics	Transcriptomics QC & batch correction · Welch DE (BH-FDR) · Olink NPX proteomics · cross-modal correlation
Longitudinal & Survival	Linear mixed-effects (lme4) · Emax PD model (nls) · Kaplan–Meier · Cox proportional-hazards
Machine Learning	PCA + UMAP · k-means clustering · elastic-net (glmnet) · random forest + OOB AUROC

Built with Quarto · R 4.4.2 · 13 May 2026

--- title: "Biomarker Data Science Platform" subtitle: "End-to-End Clinical Trial Biomarker Pipelines across Therapeutic Areas" toc: false number-sections: false --- Reproducible, modular biomarker analysis reports spanning four therapeutic areas. Each report applies an identical analytical pipeline to simulated Phase II trial data, with disease-specific biological framing and clinical interpretation. --- ## Therapeutic Area Reports :::: {.columns} ::: {.column width="48%"} ::: {.callout-note icon=false} ### [🔵 Immunology — FcRn Inhibitor](autoimmune_fcrn.qmd) **Disease** · Generalised Myasthenia Gravis (gMG) **Mechanism** · FcRn-mediated IgG catabolism (VYVGART-like) **Primary biomarker** · IgG total NPX **N** · 120 patients ::: ::: ::: {.column width="4%"} ::: ::: {.column width="48%"} ::: {.callout-warning icon=false} ### [🔴 Oncology — Checkpoint Inhibitor](oncology_checkpoint.qmd) **Disease** · Non-Small Cell Lung Cancer (NSCLC) **Mechanism** · PD-1 blockade / T-cell reinvigoration (nivolumab-like) **Primary biomarker** · Tumour Mutational Burden (TMB) **N** · 150 patients ::: ::: :::: :::: {.columns} ::: {.column width="48%"} ::: {.callout-tip icon=false} ### [🟢 Cardiovascular — PCSK9 Inhibitor](cardiovascular_pcsk9.qmd) **Disease** · Heterozygous Familial Hypercholesterolaemia (HeFH) **Mechanism** · PCSK9 inhibition / LDL-R upregulation (evolocumab-like) **Primary biomarker** · LDL Cholesterol (LDL-C) **N** · 200 patients ::: ::: ::: {.column width="4%"} ::: ::: {.column width="48%"} ::: {.callout-important icon=false} ### [🟣 Neurology — Anti-Amyloid](neurology_alzheimers.qmd) **Disease** · Early Alzheimer's Disease (MCI / mild dementia) **Mechanism** · Anti-Aβ protofibril mAb (lecanemab-like) **Primary biomarker** · Plasma p-tau181 **N** · 160 patients ::: ::: :::: --- ## Pipeline Architecture All reports share a common three-stage analytical framework executed from a single parameterised source file (`_analysis_core.qmd`): | Stage | Methods | |---|---| | **Multi-Omics** | Transcriptomics QC & batch correction · Welch DE (BH-FDR) · Olink NPX proteomics · cross-modal correlation | | **Longitudinal & Survival** | Linear mixed-effects (lme4) · Emax PD model (nls) · Kaplan–Meier · Cox proportional-hazards | | **Machine Learning** | PCA + UMAP · k-means clustering · elastic-net (glmnet) · random forest + OOB AUROC | --- ```{r} #| include: false built_str <- sprintf( "Built with [Quarto](https://quarto.org) · R %s · %s", getRversion(), format(Sys.time(), "%d %B %Y") ) ``` *`r built_str`*