N-dimensional relationships

The following images visualise the relationships between objects in N-dimensional space using the R-implementation of the SOD algorithm. The objects are biological samples or individual cells which have been characterised by gene expression measured either globally (i.e. array or RNA-seq) or for a small number of genes in individual cells by counting individual RNA molecules.

Etv2 and its direct targets

Transcripts from a dox inducible EGFP-Etv2 transcription factor and the Flk1, Fli1, Cdh5, Snail1 and Pdgfra genes were enumerated by combinatorial FISH. In addition, the amount of EGFP-Etv2 fusion protein was estimated from direct EGFP fluorescence. The transcript densities were used to determine the global set of relationships between the cells and these were reduced from 6 to 2 dimensions using the SOD algorithm. The colouring in the individual plots indicate the relative levels of either transcriptional densities or the estimated EGFP-Etv2 protein amounts.
Click image for the scalable svg file.

Effect of Etv2 induction, 2

Data was obtained as above, but plotted to indicate the transcript densities from all the enumerated genes. The area of the non-overlapping parts of the points is proportional to the transcriptional density for each gene.
Click on image for the scalable svg file.