Fully Automated Microsatellite Genotyping: Computer Software (Abstract)

See-Kiong Ng, Nandita Mukhopadhyay, Joe Rayman, Soumitra Ghosh, and Mark Perlin

Microsatellite markers have proved effective in large-scale and fine-grained genetic analysis. The full automation of microsatellite-based genotyping is currently limited by the human scoring bottleneck: it is estimated that over half the cost of microsatellite-based genotyping is due to human scoring effort. This scoring bottleneck stems largely from the PCR amplification of repetitive sequences, which introduces a stutter artifact. This artifact has thus far precluded accurate computer scoring of an individual's alleles for a dinucleotide repeat marker. Further, there is increasing interest in estimating allele frequencies for pooled DNA samples using di-, tri-, and tetranucleotide repeat markers -- these data are greatly distorted by PCR stutter.

We have developed computer-based analysis methods that automatically remove the PCR stutter artifact (Perlin, Lancia & Ng, 1995, Amer J Hum Genet, 57(5):1199-1210). These methods have been translated into a platform-independent computer program that accurately and automatically calls alleles from quantitative microsatellite data. PCR stutter and size binning information for each marker is preserved across multiple gels, thereby improving the marker calibrations. Our genotyping software has been tested thus far on ten ABI gels containing data for 5100 microsatellite genotypes. Comparison of the current software's results with human scoring shows a 3% average discrepancy rate. Our full data set in this study is 60 ABI gels, and we are working to achieve a 1% error rate. Our goal is fully automated genotyping software that can remove the human scoring bottleneck, and enable the analysis of pooled DNA samples.

in American Journal of Human Genetics, 59(4 Supplement): A229, 1996.
Back to FAST-MAP papers .
For comments, please contact perlin@cs.cmu.edu