Optimization methods for simultaneous search of multiple QTL
Sub-project of: Computational Genomics
Participants
- Sverker Holmgren (Professor in Scientific Computing)
- Carl Nettelblad Assistant Professor in Scientific Computing)
- Salman Toor (PhD in Scientific Computing)
Summary
We present an efficient and reliable multiple QTL scans algorithm for experimental populations, called PruneDIRECT. Earlier studies in our research group utilized optimization approaches, e.g. the DIRECT optimization algorithm, to QTL search problems. Speedups of several orders of magnitudes in DIRECT permits high dimensional QTL scans to be performed. However, the algorithm cannot guarantee the accuracy of the optimization process.
The new optimization scheme, PruneDIRECT, has a well-defined error bound and in practice is equivalent to a corresponding exhaustive search - exhaustive search is the popular method among QTL expertise for locating a QTL. We present simulation results that show for simultaneous mapping of three QTL using permutation testing, PruneDIRECT is typically more than 50 times faster than an exhaustive search.
We have also investigated, tentatively, the possibility to define bounds on variation in the residual sum of squares, based on bounds of the coefficients in the design matrix. This is a more general uncertainty quantification problem that seems to be understudied in this specific context.