Analysis of array/assay data utilizing transformation of scatter plot to image plot
View Patent ↗In processing images of arrays of differently-dyed particles or microbeads (“beads”), a template cluster map of the array is first generated of a first set of such beads. Each cluster is bounded and designated with an identifier. A project cluster map is then generated of an array of a different set of the differently-dyed beads. Both the template and project cluster maps are transformed into respective image plots, having regions thereof representing the density of the clusters, and the regions of the template image plot are marked for identification. The boundaries and identifiers of the regions in the template image plot are transformed to the project image plot, and the watershed algorithm is applied to the transformed boundaries in the project image plot in order to generate a set of expanded and identified regions. Finally, beads in the project cluster map are identified by correlating them with identified, expanded regions in the project image plot.
1. A method of identifying differently-dyed beads in an array of beads, comprising:
generating a template cluster map from a first array of differently-dyed beads in which beads dyed with the same dye or dyes are clustered separately from each other;
generating a project cluster map from a different array of the differently-dyed beads in which beads dyed with the same dye or dyes are are located in different regions;
forming boundaries around individual clusters in the template cluster map and designating each cluster with an identifier;
transforming the template cluster map to a template image plot, and transforming the project cluster map to a project image plot, wherein regions in the image plots are representations of the density of beads in a particular area in the cluster maps, and wherein the identifiers in the template cluster map are transferred to the corresponding regions in the template image plot;
forming a first boundary around all the clusters in the template image plot, and forming a second boundary around all the clusters in the project image plot;
establishing the positional relationship between the first and second boundaries;
forming boundaries around the individual identified regions in the template image plot;
transforming said boundaries to the project image plot using said positional relationship between the first and second boundaries, to establish a set of search boundaries;
locating the densest regions in the project image plot using the search boundaries;
applying the watershed algorithm to the densest regions in the project image plot to generate expanded, partitioned regions;
correlating the identified regions in the template image plot with their respective expanded, partitioned regions in the project image plot; and
correlating the beads in the project cluster map with the respective expanded, partitioned regions in the project image plot and thereby identifying differently-dyed beads in the project cluster map.
2. The method of claim 1 wherein the search boundaries are rectangular.
3. The method of claim 1 wherein the boundaries in the template cluster map are generated and identified by an operator.
4. The method of claim 1 wherein beads are dyed with fluorescent dyes.
5. The method of claim 1 wherein proteins or nucleic acids are coupled to the surface of the beads.
6. The method of claim 5 wherein the differently-dyed beads encode the particular protein or nucleic acid attached to the bead surface.