mlink don't like me, ilink don't like me...

I. Fenton ifenton at hgmp.mrc.ac.uk
Wed Nov 16 05:28:05 EST 1994


*shrug* i thought i was used to LINKAGE, but...

(1)  i posted a note to this group a few days ago [thanks to those who
responded] about bizarre MLINK results, namely "-infinity" results in
some families.  well, now i get some more bizarre results, but ones
that aren't as obvious to spot...  take this example of a disease v.
marker 2-pt MLINK lod score :-

theta		 lod
0%		 0.003
1%		 0.002
5%		-3.079
10%		-3.045
20%		-2.something
etc		etc

(i am making the figures up from memory, so don't quote me, but
they are roughly what happened)

soooooo, what gives ?  is this just a variation on the problem i posted
about a coupla days ago, or not ?  also is there any valid reason for
this sort of lod-score curve.  i don't think so.  this is beginnging
to get annoying.  -infinities (grammer?!) are easy to spot, but if
i'm doing a 350-marker genome scan, as i am, i could easily miss this
sort of crap.  oh dear, woe is me.

(2)  ILINK.  this is fun.  i am using ILINK to try and estimate allele
frequencies for loads of markers.  i know that i am using a 14-allele
system, so i code my DATAIN.DAT file for 14 alleles at 1/14 equal
starting values.  i tell ILINK to do the sums, but is stops after
1 iteration, and gives up.  i see that my PEDIN.DAT file has no
allele 14.  i play about, and find that if i do not have a "n" allele
in my PEDIN.DAT file, then ILINK stops, but if i change one person
from a 2-12 to a 2-14, then ILINK works luvverly, and gives me proper
frequency estimates.  why ?  there are no individuals in my PEDIN.DAT
file with either a 2 or a 7 or a 11 allele, but ILINK doesn't complain.
am i missing something stupid ?  is ILINK stupid ?  i've run a few tests,
and it always craps out early if there is no "n" allele for an "n"-allele
system.  oh dear, woe is me.

for both (1) and (2) i am using version 5.2 of LINKAGE on the pc.  i am
using a dos-pc, and will start using a lot of alcohol if i can't get
these problems fixed.  hhhhheeeeeeellllllllppppppp !!

Iain "Help" Fenton, Cardiff, UK.



More information about the Gen-link mailing list