VITESSE - Handout from 10/18/2000 course
This file briefly describes the similarities and differences in
using VITESSE compared with LINKAGE/FASTLINK.
Similarities:
- The LINKAGE utilities preplink, makeped, lsp, lrp are used in
the same way.
- The output contains similar information and is
printed both to the screen and to files.
Differences:
- VITESSE provides the functionality of both MLINK and LINKMAP
with one program.
- The LINKAGE/FASTLINK pre-processor program UNKNOWN is not used.
- The scripts produced by lcp must be passed through a filter called
cnvrt_sh to make them suitable for VITESSE.
- The output contains a lot of diagnostic information about how the
computation is proceeding. This diagnostic information is
helpful only if you read the paper about VITESSE.
- VITESSE computes exact likelihoods, while LINKAGE/FASTLINK compute
scaled likelihoods that are multiplied by a number that depends on the
pedigree structure. The effect of the scaling is eliminated when
one takes the difference of two log(likelihood) terms as is done
in computing a lodscore.
- The standard output file names are:
LINKAGE/FASTLINK VITESSE
final.out vfinal.out
stream.out vstream.out
Here are some brief examples of how to run VITESSE.
We start with shell scripts for MLINK and LINKMAP as produced
by lcp (see the instructions for lcp). Then run:
(for MLINK)
oceanus% cnvrt_sh
Input file :mlink-pedin
Output file: vmlink-pedin
-> 88 lines processed
oceanus% vmlink-pedin
(for LINKMAP)
oceanus% cnvrt_sh
Input file :linkmap-pedin
Output file: vlinkmap-pedin
-> 155 lines processed
oceanus% vlinkmap-pedin
If VITESSE sees a complex (i.e., not simple) pedigree in the
input. It will abort. If you have a data set with multiple
pedigrees, some of which are not simple, my advice is:
- Use FASTLINK MLINK
- Use a combination of FASTLINK LINKMAP for the complex pedigrees and
VITESSE for the simple pedigrees. Then use
Jim Tomlin's (jtomlin@helix.nih.gov) FASTESSE perl program
to automagically merge the FASTLINK and VITESSE results.