Effect of transmission model specification on lod score analysis (2)
If the transmission model specified for lod score analysis differs from
the true model then the lod scores obtained will be different from those
the true model would have produced. The nature of the differences is
somewhat random and unpredictable, but certain general rules apply.
If the penetrance is set too high and/or the phenocopy probability too
low and/or the disease mutation frequency too low (i.e. the genetic
effect is set too large):
- The maximum lod score obtained close to the disease locus will be
somewhat reduced
- The recombination fraction at which the maximum lod score is
obtained will be inflated
- The lod score at small recombination fractions may be strongly
negative, even for markers tightly linked to the disease locus
In general the lod score curves produced will be more sharply curved and
have a greater magnitude.
By contrast, if the penetrance is set too low and/or the phenocopy
probability too high and/or the disease mutation frequency too high
(i.e. the genetic effect is set too small) then flatter lod score curves
of a smaller magnitude tend to be produced.
If the transmission model is grossly misspecified (e.g. recessive
instead of dominant) then lod scores may be completely negative, even
with linked markers.
modspec1.html, modspec3.html