As of SVS version 6.3, Golden Helix PBAT includes an option when doing family-based SNP analysis to use a new Alternative Rapid Extended Pedigree Algorithm (available under the Phenotype and Haplotype Parameters tab). The greatest benefit of this new algorithm is that it's significantly faster than traditional PBAT analysis on extended pedigrees. At this time we label it as experimental because you may get inconsistent results between the two options. Included here is an abbreviated description of what the algorithms do "behind the scenes" and why you may experience different results.
----------
The new algorithm for extended pedigrees combines the advantages of two strategies:
- Breaking up extended pedigrees into nuclear families. This is a computationally fast strategy, but it does not take full advantage of the known extended pedigrees structures.
- Analyzing extended pedigrees as such. This takes full advantage of all the information and is the most powerful option, but can be computationally slow when a lot of the genotypes in a pedigree are missing.
The traditional extended pedigree algorithm is particularly slow in situations in which nuclear families (i.e. all genotypes are known) within an extended pedigree can only be linked by two or more family members without genotypes or extended pedigrees with "isolated genotypes" (i.e. spare genotypic information spread across the entire pedigree). In these situations, the power gain is minimal and sometimes even jeopardized by the possibility that a family-member has to be removed when the maximum number of founders is reached in PBAT.
The new algorithm for extended pedigrees in PBAT identifies clusters of nuclear families within extended pedigrees that are directly linked (i.e. share a family-member) and analyzes such clusters as extended pedigrees. At the same time, clusters that can only be linked through two or more family members without genetic information are broken up into "extended pedigrees". These extended pedigrees are analyzed as extended pedigrees, but as if they were independent of each other. The minimal information that such extended pedigrees provided was in the computation of the genetic distribution algorithm and served only as a disadvantage by increasing the computation time un-proportionally.Such links between clusters of extended pedigrees are therefore dropped in the new hybrid-approach. The new approach maintains the increased statistical power of the original extended pedigree algorithm, while having almost the computational speed of a pure nuclear family-analysis.
