Search

Most flow/image cytometric data analysis methods look for clusters in the data corresponding to specific cell subpopulations. Comparisons between different cytometry datafiles often use human pattern recognition visualization of all the different combinations of variables ("parameters") two at a time in so-called bivariate scattergrams. Not only is this tedious, but it can miss potential clusters due to projection of higher dimensional dataspaces down onto two dimensional planes making them indiscernible as separate clusters. Novel data mining algorithms, implemented in software allow for the comparison of two or more higher dimensional datafiles without the requirement for reduction of dimensionality for human visualization. Of equal importance is the comparison of higher dimensional clusters which may move around slightly in space, yet still be "similar" according to algorithms which provide measures of similarity. This software, written in C/C++ and currently implemented in software with a Windows graphical user interface, allows for direct reading of FCS2.0 format flow cytometry datafiles of any number of parameters. In a few minutes or less, complex multiparameter data of two or more files can be compared on a personal computer or workstation. The software operates in either supervised or unsupervised mode, depending on whether the user wishes to include prior user knowledge or in a data mining discovery mode. Differences between these files can be exported as sub-datafiles which can be further analyzed using any other software that can read FCS2.0 data format.

A novel data mining program called “subtractive clustering” picks out the most important differences between two or more flow cytometry listmode data files. While making no assumptions about the data, the program uses a variable weight and skew metric in the determination of bin size allowing for subtractive clustering of data without the need for bit-reduction or projection. In contrast, other subtraction methods, such as channel-by-channel subtraction, are dependent upon dimensionality and resolution, which can lead to an overestimation of positive cells because they do not account for the overall distribution of the test and control data sets. By taking into account human visual inspection of the data it is possible for the experimenter to choose an optimal subtraction by choosing an appropriate weight and skew metric, but without allowing direct modification of the results. By maximizing a bin size which can still differentiate clusters, it is possible to minimize computation while still removing data. The choice of control weight allows for different levels of bin destruction during the subtraction stage, the smaller the number the more conservative the subtraction, the larger, the more liberal. Three data sets illustrate full dimensional subtraction, single step biological data and multi-stage subtraction to show definitive test results. Subtractive clustering was able to conservatively remove control information leaving populations of interest. Subtractive clustering provides a powerful comparison of clusters and is a first step for finding non-obvious (hidden) differences and minimizing human prejudice during the analysis.

Gene expression microarray analyses of mixtures of cells approximate a weighted average of the gene expression profiles (GEPs) of each cell type according to its relative abundance in the overall cell sample being analyzed. If the targeted subpopulation of cells is in the minority, or the expected perturbations are marginal, then such changes will be masked by the GEP of the normal/unaffected cells. We show that the GEP of a minor cell subpopulation is often lost when that cell subpopulation is of a frequency less than 30 percent. The GEP is almost always masked by the other cell subpopulations when that frequency drops to 10 percent or less. On the basis of these results one should always assume that the GEP of a given cell subpopulation is probably seriously affected by, the presence of significant numbers of other "contaminating" cell types. Several methodologies can be employed to enrich the target cells submitted for microarray analyses. These include magnetic sorting and laser capture microdissection. If a cell subpopulation of interest is small, very high-throughput cell separation technologies are needed to separate enough cells for conventional microarrays. However, high-throughput flow cytometry/cell sorting overcomes many restrictions of experimental enrichment conditions. This technology can also be used to sort smaller numbers of cells of specific cell subpopulations and subsequently amplify their mRNAs before microarray analyses. When purification techniques are applied to unfixed samples, the potential for changes in gene levels during the process of collection is an additional concern. Since RNA rapidly degrades, and specific mRNAs turn over in minutes or hours, the cell separation process must be very rapid. Hence, high-throughput cell separation (HTS) technologies are needed that can process the necessary number of cells expeditiously in order to avoid such uncontrolled changes in the target cells GEP. In cases where even the use of HTS yields only a small number of cells, the mRNAs (after reverse transcription to cDNA's) must be amplified to yield enough material for conventional microarray analyses. However, the problem of using "microamplification" PCR methods to expand the amount of cDNAs (from mRNAs) is that it is very difficult to amplify equally all of the mRNAs. Unequal amplification leads to a distorted gene expression profile on the microarray. Linear amplifications is difficult to achieve. Unfortunately, present-day gene-chips need to be about 100 times more sensitive than they are now to be able to do many biologically and biomedically meaningful experiments and clinical tests.

A number of theoretical and practical limits of high-speed flow cytometry/cell sorting are important for clinical diagnostics and therapeutics. Three applications include: (1) stem cell isolation with tumor purging for minimal residual disease monitoring and treatment, (2) identification and isolation of human fetal cells from maternal blood for prenatal diagnostics and in-vitro therapeutics, and (3) high-speed library screening for recombinant vaccine production against unknown pathogens.