Uwe Ohler's Research Group

Image expression analysis

Image analysis is a relatively young sub-field in computational biology, but has become increasingly mainstream with the arrival of high-throughput microscopy data. We are particularly interested in such data as "high-resolution gene expression data", i.e. to use it to obtain and utilize information on the spatiotemporal expression of genes.

Expresion levels from confocal images.

Dan Mace has developed a prototype to determine tissue-specific expression levels from confocal microscopy images. Our current dataset shows the expression of Arabidopsis transcription factors important for root development. Look at example pictures for a first impression of this work. Contains many beautiful pictures.

Comparing spatial expression profiles.

A second ongoing project deals with the issue of appropriate distance metrics to compare spatial expression patterns. For this, we make use of Drosophila embryo RNA in situ hybridization images. Our registration approach based on shape models came out in 2009, and more recently, we have used sparse Bayesian factor models to automatically decompose and classify the extracted spatial expression patterns. You can find supplementary data as well as a java tool implenenting our registration approach.