This image is one of my favorite scientific visualizations and it makes for a great large scale print. This image only exists in computational space and represents a glancing section of real retina from a mus musculus probed with antibodies to three small molecular labels from Signature Immunologics. Anti-taurine, anti-glutamine and anti-glutamate assigned to red, green and blue color channels respectively. This image shows two fundamentally important things about metabolomics: 1) Diversity of metabolomic signatures across cell classes is extensive and in this case, it shows how complex the retina is. 2) Within cell classes, metabolomic envelopes are very narrowly constrained in healthy tissue. Pathology is another issue for another time…
This image was analyzed using tools called Computational Molecular Phenotyping (CMP) that reveal the metabolic state of the all cell types in tissues.
The whole point behind CMP is that we can do this in N-space or use N labels to look at co-segregation of small molecular signals. We commonly look at 7-12 probes at once for instance, but they are difficult to show in traditional rgb images. The use of advanced clustering approaches is important because with the image above, we had 9 separate probes that we used on this dataset. That means, according to this formula (n+r-1)!/r!(n-1)!, there are 165 different ways to combine probes into 3 color datasets for visualization. With 12 separate probes, that is 364 different combinations and 24 separate probes opens this up to 2,600 different potential 3-color combinations.
The individual labeled images are below in grayscale along with a couple of two color composites. Note that these are just 3 color images.
Taurine in red, Glutamine in green.
Taurine in red, glutamate in blue.