What is FuSiOn?

The Functional Signature Ontology or FuSiOn in short is an ontology map built on functional signature of human kinome and miRNAs. The functional signature was generated by 6 probe gene expression profiles measured in high throughput bead based multiplex assay platform in response to different genetic and chemical perturbations including siRNAs, miRNAs, and natural products. siRNA is a chemically synthesized double stranded RNA, when transfected, destabilizes the complementary mRNA or inhibits it's translation through the RNA interference mechanism. miRNA is a group of noncoding RNAs encoded in a human genome, when expressed endogenously or transfected, regulates target gene expressions post-transcriptionally like siRNA. These functional genetic signatures gain access to chemical space by profiling unique natural products with the same assay platform. Introduction to John MacMillan's natural products can be found at http://www4.utsouthwestern.edu/macmillanlab/research.html.

What can be done?

FuSiOn can be used to search for genetic or chemical matters that behave functionally similarly to your query gene. Here are some examples.

How was the FuSiOn built?

Assay:The 6 probe genes (ALDOC, NDRG1, BNIP3L, BNIP3, ACSL5, LOXL2) were originally developed by our group in collaboration with the Robert Lewis's lab at University of Nebraska (http://www.unmc.edu/eppley/lewislab/). Variable signals from the probe genes are used to differentiate distinct genetic/chemical perturbations. Using Luminex HTG , expression of the 6 genes as well as two control genes, HPRT and PPIB, was measured in one-perturbation/one-well based 96 well plate format after 48 (?) hours of transfection of 780 siRNAs and 345 miRNAs or addition of 1200 different fractions of natural product provided by John MacMillan at UT Southwestern medical center.
Data processing:After correcting background noise, 6 gene expression values/perturbation were geo-mean normalized to the two internal control probes, and the results were further probe median normalized and log transformed to smooth signal level between the probes. There are relatively silent perturbations not signaled by any one of the six probes. After censoring those by IQR statistics using the noise model from the negative control wells, a total of 300 kinases and 100 miRNAs serve as a high confidence subset for the further analysis as well as the full set. All the 720K pair-wise distances were measured between the genetic perturbations, and between them and all the 1200 natural product fractions using euclidean, mahalanobis distance, and pearson correlation. Statistical significance of the distance is estimated from the permuted dataset. Distance sum or rank sum over triplicates was used to estimate the empirical probability from the permuted dataset and to estimate the false discovery rate. Further details can be found at http://stke.sciencemag.org/content/6/297/ra90.full.