K. B. TRANG

From a patient’s concern to a druggable target.

Khanh B. Trang, MD, MPH, PhD · Children’s Hospital of Philadelphia. I join human genetics and multiomics to find which genes, in which cells, are worth drugging.

01 Population a population

Genome-wide studies find thousands of risk variants. Association alone is not mechanism.

Autoimmune study16 autoimmune GWAS

Read the evidence → Scroll to zoom · hover the rings for detail Scenes are illustrative

Genetics that ends in a cell type. Atlases that end in a target.

Who
Clinical scientist, geneticist, epidemiologist (MPH) and computational biologist (PhD, Marie Skłodowska-Curie fellow). Senior Staff Bioinformatics Scientist, Division of Human Genetics, Children’s Hospital of Philadelphia.
What I join
◆ GWAS, fine-mapping, heritability, Mendelian randomisation × ● 3D chromatin, single-cell, spatial, CRISPRi screens
Proof
  • 57 cell types × 16 autoimmune diseases → FDFT1, and an existing inhibitor that shifted T-cell cytokines. Genome Biology 2025 →
  • 89 GWAS-prioritised elements targeted by CRISPRi: 23 genes responded, chromatin loops alone had predicted 8, and 4 held up across three follow-up assays. Genome Biology 2025 →
Heritability of 16 autoimmune diseases partitioned across cis-regulatory elements of 57 cell types (S-LDSC); regulatory T cells lead for type 1 diabetes.Autoimmunity · 57 cell types The CRISPRi screen: target origins, test statistics and responsive genes, with loci where silencing changed ARID5B and NCOA3.Bone · CRISPRi screen A single-nucleus atlas of hepatoblastoma and adjacent liver: cell types and their composition across samples.Hepatoblastoma · single-nucleus atlas Putative childhood-obesity effector genes across cell types, including the ADCY3 locus.Childhood obesity · effector genes
Narration transcript
  1. Genome-wide studies find thousands of risk variants. But association alone does not tell you the mechanism.
  2. Behind every GWAS signal is a person with a genetic predisposition, and disease can arise in one organ or in several.
  3. Disease happens in tissue, and cells don't act alone. Spatial omics maps each organ's cells within their microenvironment, or niche, from the layers of the brain to the follicles of a lymph node.
  4. Grouped by gene expression, cells form types that behave differently across disease states and stages. Among fifty-seven cell types, regulatory T cells showed the strongest enrichment of type 1 diabetes genetic risk.
  5. Placed back in each tissue, cell types show which neighbourhoods they share.
  6. Folded chromatin brings regulatory elements into contact with genes, nominating candidates, including distant ones.
  7. Most risk variants sit in cis-regulatory elements, such as enhancers, that loop across the chromatin to reach the genes they control.
  8. CRISPR screens then test which predicted links actually change a gene. In one screen, chromatin loops had predicted only eight of the twenty-three genes that responded.
  9. Linked genes then meet existing drugs. For example, an inhibitor of FDFT1 shifted T-cell cytokines: a pharmacological hypothesis worth testing further.
  10. From a patient's concern to a hypothesis worth testing. This is the work of Khanh B. Trang.

Narration uses a synthetic voice generated for this site.

How I integrate →All work →Email me