jarny.net
Jarny Choi
Research from beneath the surface: from tissues and cells down to a single letter of DNA.
1 mm · tissue · former work
Atlas buildingWhere everything was, once.
Reference atlases bring many datasets together into one shared map of cell types and states, so that a new sample can be placed against everything that came before it.
-
Stemformatics data portal enables transcriptional benchmarking of lab-derived myeloid cells
How to match lab-derived myeloid cells against integrated reference atlases of human macrophages and dendritic cells.
-
Haemopedia RNA-seq: a database of gene expression during haematopoiesis in mice and humans
RNA-seq profiles of 57 mouse and 12 human blood cell types, explorable through the Haemosphere portal.
-
A simple, scalable approach to building a cross-platform transcriptome atlas
850 samples from 38 platforms combined into one atlas of human blood cells, by keeping only genes with low platform bias.
10 µm · cell · former work
Stem cellsBefore it decides what to become.
Earlier work on stem cells: how a cell keeps many possible futures open, and what its molecular state says about which one it will choose.
-
In vivo clonal tracking reveals evidence of haemangioblast and haematomesoblast contribution to yolk sac haematopoiesis
Single-cell sequencing and cellular barcoding trace yolk sac blood, endothelium and mesenchyme back to three distinct precursors, one of them newly described.
-
Transcriptional profiling of stem cells: moving from descriptive to predictive paradigms
A review of how expression data can move from describing stem cells to predicting their identity, and the data traps along the way.
-
Age-specific biological and molecular profiling distinguishes paediatric from adult acute myeloid leukaemias
The age of the cell of origin shapes acute myeloid leukaemia, with young AML marked by upregulated immune pathways.
100 nm · chromatin · most papers since 2023
Method developmentInstruments for seeing in the dark.
Most of my papers in the past three years are about new methods: computational tools that make structure visible in data where it would otherwise stay hidden.
-
Scalable cell-specific coexpression networks for granular regulatory pattern discovery with NeighbourNet
A co-expression network for every individual cell rather than every cluster, to catch regulation that shifts gradually between cells.
-
Φ-Space ST: a platform-agnostic method to identify cell states in spatial transcriptomics studies
Continuous cell states in spatial transcriptomics data from any platform, using several single-cell references at once.
-
StableMate: a statistical method to select stable predictors in omics data
Separates predictors that hold across datasets from those that only hold in one batch or environment.
1 nm · a single variant · current work
eQTL analysisOne letter, out of place.
Most genetic variants linked to disease lie outside genes, in DNA that doesn't code for protein. eQTL analysis asks which of those variants change how strongly a nearby gene is expressed, and where.
-
Flexible and efficient count-distribution and mixed-model methods for eQTL mapping with quasar
Fast, flexible software for eQTL mapping, from linear models to count-based mixed models.