Plant Genome III Conference Plant Genome III Conference
Conventional statistical software is not well adapted to viewing, and analyzing molecular-marker and phenotypic data collected on plant mapping populations nor to the application of such data to marker- assisted selection (MAS). Some disadvantages are: generality, necessity of programming, poverty of graphics. I describe working software that addresses these problems. Written in C and running on a Macintosh computer, it accepts marker and phenotypic datasets for a standard mapping population (RI, BC1, F2, DH) and a map containing the markers. Using a menu and dialog interface, it quickly performs simple and interval-based (Haley & Knott, Heredity 69: 315-324, 1992) regressions of phenotype on marker genotype and produces graphical plots of the resulting statistics along chromosomes. Statistics are exportable as text and include F-tests, P-values, R2 etc. Graphics are also exportable. Other analyses, also combining graphics and numerical output, include frequency histograms for traits, segregation of marker loci, molecular similarity among plant genotypes in the population, two- way locus interactions, and multiple regression of phenotype on loci and their interactions. Features in preparation include the analysis of nonstandard populations (advanced backcross) and genetic systems (autopolyploid), use of more appropriate statistical models, construction and application of selection indices for MAS, and QTL inference based on combination of phenotypic data from multiple trials. The efficiency of QTL analysis by software specifically designed for the task suggests its incorporation in plant genome databases, perhaps as externally supplied code routines. This would be one way to standardize QTL description.