Applied Biosystems 7900HT User Manual page 165

Algorithmic
Manipulation of
Raw Allelic
Discrimination
Data
About the
AutoCalling
System
Applied Biosystems 7900HT Fast Real-Time PCR System and SDS Enterprise Database User Guide
Table 5-1 Signal and Sequence Correlation
A substantial increase in...
both fluorescent signals
The SDS software can analyze raw data immediately upon completion of an allelic
discrimination run. The term 'raw data' refers to the spectral data between 500 nm to
660 nm collected by the SDS software during the plate-read. During the analysis, the
software employs several mathematical algorithms to generate from the raw data a more
direct measure of the relationship between the spectra changes in the unknown samples.
The first mathematical algorithm involves the conversion of the raw data, expressed in
terms of Fluorescent Signal vs. Wavelength, to pure dye components using the extracted
pure dye standards. After the software identifies the dye components, it determines the
contribution of each dye in the raw data using the multicomponent algorithm.
After normalization, the software processes all samples associated with markers that are
configured for AutoCalling using the Applied Biosystems proprietary Maximum
Likelihood Algorithm. The algorithm conducts a cluster analysis of the data based on the
ratio of normalized reporter dye signal (for example, the ratio of FAM
The result of the analysis typically yields three major clusters corresponding to the three
genotypic constituents: Allele X homozygous, Allele Y homozygous, and heterozygous.
After clustering the data points, the software applies a series of probability tests to
further refine the clustered data and to identify outliers. For every sample, the software
calculates a final "quality value" that represents a measure of the likelihood that the
sample belongs to each cluster. The algorithm then applies the quality value for the
marker (set in the Analysis Settings dialog box) as a threshold for calling the associated
sample data. Sample-cluster associations that generate quality values greater than the
defined threshold are automatically assigned the call of the associated cluster.
Any SNPs which depart substantially from expected Hardy-Weinberg
Note:
frequencies should be reviewed.
About the 2-Cluster Calling Feature
If only two clusters are present in a study, the AutoCalling algorithm uses the expected
genotype frequencies of the clusters to assign their genotypes. If a study contains two
large clusters and a single sample between the two clusters, the software considers the
intermediate sample as an outlier rather than a cluster, and treats the data as a two-cluster
study.
Note: The Maximum Likelihood Algorithm is designed to use No Template
Controls, and autocalling is most effective when NTCs are tasked.
Indicates...
heterozygosity.
DRAFT
September 1, 2004 11:39 am, CH_End-Point.fm
Overview
™ to VIC ® dye).
5-7