Multipoint human genetic linkage analysis can be extremely demanding of computer resources. Problems requiring hundreds of hours of computer time per point are not uncommon.
The aim of this research is to investigate the use of high performance computers (generically referred to as supercomputers) on the linkage mapping problems. We propose to design software to test mapping algorithms on two types of computers. One is a high performance vector computer (a Cray YM-P machine). The second is a highly parallel scalar computer of the type currently entering the research arena, a 64 processor BBN Butterfly GP1000. The programs necessary to maximize performance on these two machines are very different. A comparison should provide a good indication of which architecture is most suitable for this type of problem. This work should also result in linkage mapping programs that execute many times faster than those currently available.
Showing the most recent 10 out of 97 publications