Simple intraplex concept diagram showing idealized characteristics. A. m = 5 different microsphere sets (i.e. 5 SMPCS's) (labeled 01 to 05) are shown. Their respective coatings of ligand (in this case antigen) to bind analyte (in this case antibody) are varied by consecutive dilutions. So, more binding sites are available for a target antibody analyte on those microspheres incubated with higher concentrations. B. Reporter fluorescence readings for an assay that reflect the 2× series dilutions of ligand bound to microspheres showing how each set responds differently to the same concentration of analyte. Mean of m = 6200 as denoted by horizontal line. This is the internal self mean of the m fluorescence readings. C. Internal self mean ratios for each of the SMPCS's. Example calculation shown for SMPCS 01. The mean of m is used as the denominator for each of the m fluorescence readings.

Histogram of intensities of reporter fluorophore for microsphere classifier set #97, that has an N of 136. This is a representative sample of the histograms generated by extracting event data from the Bio-Rad Bioplex XML data file. Visual inspection shows a fairly normal distribution with high end outliers in a long tail.

Mean inter-instrument ratio Instrument A/Instrument B. This shows that two different instruments, both under standard service contracts, will not necessarily have the same responses for all concentrations, despite being calibrated using the same standard. This suggests that there is potentially significant variance in the response curves of the parts making up the opto-electronic system. However, the intraplex method eliminated this and other problems.

SRV assay external ratios (Y axis) using mean for a different assay set as denominator (SRV/SFV mean). (Corresponds to Table 2 B.)This ratio appears to work better than that shown in figure 6, which is attributed to apparent greater variance in the uncoated sets than is seen in real assays. Ranges for three different concentrations of serum are shown, and it is possible to see how ratios cluster closer together as concentration of serum goes down. Compare with figures 6 and 8.

Ratios on internal self mean of set. (Instrument A/Instrument B). As can be seen here, a ratio on the internal self mean gives good correlations between instruments for all three intraplex assays. There is some separation at lower concentrations, which is expected as the signal to noise ratio declines. (Corresponds to Table 2 A.)

Mean and standard deviation by type of ratio taken. This graph shows the mean inter-instrument comparison of the ratios by type, and their standard deviations by type. What one looks for here is a mean ratio that is closest to one, combined with the smallest standard deviation. In this graph is seen summarized the data seen in different form in figures 6, 7 and 8, respectively for the three items in this graph.

Concept diagram for m × n microsphere matrix. A. Each circle in this diagram represents a set of microspheres (i.e. an SMPCS). Each of the superset identical groups (i.e. SMPCS-IDG) (m = 5) of are coated at different sensitivities. The SMPCS-IDG's of m are across the top, labeled 01–05. Note that now each m is a superset composed of 5 microsphere set identifiers (i.e. an SMPCS-IDG). Each of n (01 to 05 for SMPCS-IDG 01, 06 to 10 for SMPCS-IDG 02, ...) microspheres that make up the superset SMPCS-IDG for m is coated in the same batch for identical sensitivity. Like figure 1, the m SMPCS-IDG's have serial dilutions (or some other useful difference in sensitivity method) in their manufacture. B. Processing of the intraplex using a simulated example. Step 1: On left is an m = 5 × n = 5 fluorescent reporter reading dataset graph for all SMPCS's, 01–25. (Note the outlier at 05, S5 that was removed for the set of n for the m SMPCS-IDG number 01.) Completion of step 1 is removal of outliers. Step 2: The result of this step is m averages, (means of n) using as input the n microsphere set fluorescence readings for each SMPCS-IDG. This is shown in the table. Each of these 5 means of n are averaged together to give a single mean of m. Step 3: Internal self mean ratios using the mean of m as denominator for each of the means of n from step 2. This is done in the same way as for the simple intraplex of figure 1.

Ratio of trimmed mean/median. For this study, it was useful to use trimmed mean so that standard deviations would be available for each reading. This graph shows that the trimmed mean is close to the median which is commonly used for this instrument. This is also a strong indication of normal distribution. Y axis is mean fluorescent intensity (MFI).

SRV assay external ratios (Y axis) using mean of uncoated microspheres as denominator. (SRV/uncoated mean). (Not used in Table 2.) This is one of two external ratios that were taken. Uncoated microspheres were one of three controls in the experiments, and one of two controls that had multiple SMPCS's. Ranges for three different concentrations of serum are shown, and it is possible to see how ratios cluster closer together as concentration of serum goes down. Compare with figures 7 and 8.

SFV assay internal ratios using internal self mean as denominator (SRV/SRV mean). (Corresponds to Table 2 A.) This ratio conveniently turned out to be the most effective at controlling for all types of variances. Like figures 6 and 7, ranges for three different concentrations of serum are shown, and it is possible to see how ratios cluster closer together as concentration of serum goes down. Compare with figures 6 and 7.

External ratios on uncoated mean. (Instrument A/Instrument B) This figure shows ratios on an external mean, where an external mean is the mean of an assay for a different analyte. This graph demonstrates that, on average, an apparently quite stable external mean is not as good as an internal mean ratio. Comparing Figures 9 and 10, one can see that Figure 9 has ratios that are closer to the desired ratio of 1. (Corresponds to Table 2 B.)