04 Dewpoint Therapeutics · Multimodal analysis
Condensate profiling
At Dewpoint Therapeutics, I combined RNA sequencing and high-content microscopy in a liver cell model of type 2 diabetes (T2D). The goal was to identify potential new drug targets by looking for changes in biomolecular condensates associated with insulin resistance.
Motivation
Look beyond the insulin receptor
Biomolecular condensates are dynamic assemblies that concentrate proteins and other molecules within cells. Research on insulin resistance found that insulin stimulation recruits fewer insulin receptors into clusters in insulin-resistant cells. Dall’Agnese et al., 2022
I asked whether other condensates involved in liver metabolism also changed under conditions that model T2D, and whether those changes coincided with shifts in gene expression.
Quantitative approach
Compare matched measurements
The experiments varied glucose and lipid levels in human liver cells, with and without insulin stimulation. I applied a variance-stabilizing transformation to gene-level counts and standardized microscopy features such as condensate size, count, and intensity. I then aligned the two data types by sample and fit Multi-Omics Factor Analysis (MOFA2).
Workflow
Separate shared and modality-specific patterns
Some factors mainly reflected one data type. One factor linked shifts in gene expression to changes in the organization of condensates involving transcription factors. I used the feature weights to see which genes and imaging measurements contributed most. Those associations can help prioritize which condensates to test as potential drug targets.