Linkage Newsletter
Vol. 6 No. 2 July 1992
Published by Jürg Ott, Columbia University, New York.
-
Editorial Assistant: Katherine Montague
-
Fax: +1-212-568-2750
Tel. 212-960-2507
e-mail: OTT@NYSPI.BITNET
-
Postal address:
- Columbia University, Unit 58
722 West 168th Street, New York, NY 10032
EDITORIAL
This summer issue of the newsletter is rather short. It is
again being distributed via e- mail (to the GENETIC-LINKAGE
newsgroup of the BIOSCI network and to individual subscribers) as
well as by postal mail. Hope everyone has a nice summer!
LINKAGE COURSES
The following linkage courses are scheduled:
Zurich (Advanced course): October 19-23, 1992, in the
Computer Center at the University of Zurich, Irchel campus (see
full-page course announcement attached).
New York (Advanced course): January 11-15, 1993, in the
microcomputer classroom of the Health Sciences library, Columbia
University. Course fee $100 (supported by National Center for
Human Genome Research; only a small number of participants from
outside the U.S. can be admitted). Registration is open; topics
covered are as in the Zurich course. A formal course announcement
will be made later.
In the spring of 1993, there will also be introductory
courses in New York and Zurich but dates have not yet been fixed.
For information and application forms for any course, please write
(preferably by fax) to Katherine Montague, course coordinator, at
the address given above.
SOFTWARE NOTES
Bug in Vax version of MLINK?
Joseph Terwilliger recently noticed the following
discrepancy between the Vax version and the PC version of the MLINK
program. Consider two parents and their child who are typed for
two 2-allelic markers. Both parents are heterozygous, 1/2, at each
locus and the child is homozygous, 1/1, at each locus. There is
complete linkage disequilibrium, P(1-1) = P(2-2) = 0.5, and P(1-2)
= P(2-1) = 0. Without allelic association, of course, this family
is uninformative for linkage. Under complete disequilibrium,
however, given the offspring's phenotype, parental phases are known
and lead to two nonrecombinations. Thus, the lod score is given by
Z(é) = log[4(1-é)2], which has a maximum of 0.60 at é=0. This is
also the result furnished by the PC version of MLINK, but the Vax
version reports zero lod scores at any é. We are investigating
this matter and will report results as soon as possible.
Undeclared loops
In the LINKAGE programs, when loops (marriage or
consanguinity) are present in a pedigree, this must be specified in
the pedigree file (the safest method is to use the MAKEPED
program). If a loop remains undeclared, depending on the type of
loop, the analysis programs may terminate with an error or, worse,
they may appear to terminate normally but give incorrect results.
Xiaoli Xie has now written a program to detect loops (Xie X, Ott J
[1992] Am J Hum Genet, abstr, in press). It is best run after
MAKEPED in which case it catches any loops the user failed to
declare. This LOOP program is based on the depth-first search
algorithm in graph theory and is freely available.
Simulating under heterogeneity in SLINK
Both the SIMLINK (M. Boehnke) and SLINK (D. Weeks) programs
allow for simulating pedigree data under heterogeneity, that is,
with a given proportion of families simulated without linkage
between trait locus and marker loci. In the analysis, however,
there is a major difference between the two programs: The SIMLINK
program analyses the data under heterogeneity while the analysis
programs of the SLINK package do not. Some users have previously
been unaware of this and have analyzed the data under homogeneity
even though they had been generated by SLINK under heterogeneity
(the resulting expected lod scores are too small). The program
ELODHET was developed to allow analysis under heterogeneity for
data generated by the SLINK program. It is part of the current
package of SLINK.
Change in the LINKLODS program
The LINKLODS program, which comes with the PC version of
LINKAGE, calculates lod scores for individual families from the
output of MLINK or LINKMAP. Dr. Chantal M‚rette recently pointed
out that occasionally the LINKLODS program gives incorrect "total"
lod scores. This occurred when a large number of families was
analyzed and the total log likelihood was smaller than the program
constant lowlod, which previously was set to -500. This constant
has now been changed to a value -10000 so that the error should no
longer occur. The current program version is 1.70 (6 July 1992).
Linkage analysis with highly polymorphic markers
Large numbers of alleles can pose problems in linkage
analysis. Various exact and approximate ways of overcoming these
problems have been proposed. Before discussing an overview of
these possibilities, I would like to ask the readers if anyone has
experience with the URP program (Michael S. Braverman: "An
algorithm to improve the computational efficiency of genetic
linkage analysis," Comp Biomed Res 18, 24-36, 1985). Please let me
know Ä I'm sure many researchers will be interested in this topic.
CORRECTIONS IN ANALYSIS OF HUMAN GENETIC LINKAGE
Below, the currently known corrections to this book (J. Ott,
1991, Johns Hopkins University Press, Baltimore) are listed.
Page 14, line 4 up: Assumption (2) is sufficient for that
statement; (2) implies (1).
Page 18, line 8: Replace (1.3) by (1.2).
Page 38, Problem 2.2: Replace 200 cM by 100 cM.
Page 44, line 8 below table 3.1 should read: "Generally,
for phase known data, if T=k/n is the value of...". Also, line 12
should read: "Since T is unbiased, ..."
Page 47, lines 5-8: These two sentences are clearer when
worded as follows: "Consider now our previous hypothetical example
of one recombinant and four nonrecombinants and test H0:é=« against
H1:é=0.1. For these data, the likelihood ratio is calculated as
Tobs = [0.1*(0.9)^4]/(0.5)^5."
Page 48, line 16: Replace A÷(1-á) by A÷(1-á)/à.
Page 59, line 3 from the bottom: ..., P(e) = 1/22, ...
Page 60, lines 6 and 7 should read: The ith segment (i =
1..s), of length bi, then contains the likelihood ratio, L*(éi),
where b1 = «(é2+é1), bi = «(éi+1+éi)-«(éi+éi-1) = «(éi+1-éi-1), bs
= 0.5 - «(és+és-1); äbi = 0.5.
Line 17 should read: 52.672, resulting in a value of 0.71
for Smith's (1959) posterior...
Table 4.1: The values of bi for i=1 (now 0.025) and i=2
(now 0.050) should be 0.030 and 0.045, respectively. This way,
they are consistent with the definition of the bi's further up on
page 60.
Page 34, lines 17 and 18 up are clearer when formulated as
follows: "... often used before linkage analysis as a preliminary
test of paternity."
Page 45, lines 12 and 13 should be phrased more exactly as
follows: "..., which allows the calculation of approximate
confidence intervals from asymptotic variances... ."
Page 68, last line before section 4.5: Replace 11.7 by
11.6.
Page 74, line 3: Replace Z() and Z(f) by Z1(m) and Z2(f).
Page 75, line 5: Replace (1-à1)n by (1-à1)g.
Page 92, line 3: Replace A1 by A2.
Page 93, table 5.3, line i=4: Replace AB-22 by AB-11.
Page 101, after equation (5.15): Replace 1/[n x i(r)] by
1/[n x i(r)]«.
Page 101, line 6 in section 5.9 should read: "type 1 is a
recombinant under one of the parental phases (phase I, say) but a
nonrecombinant under the other, ..."
Page 117, lines 21-23: The last sentence in this paragraph
should read: The second child has genotype 121/222 or 122/221,
each of which requires at least one recombination in the father or
the mother.
Page 137, first line, should read: ...between the loci C and
D.
Page 139, Table 6.10, line R: Replace "444éBC" by "444éAB".
Page 148, line 11: Replace fdd by fDD.
Page 149, table 7.1, line d1/d1: replace « by «r for P(g;r)
(as on the line above it).
Page 216, Problem 9.2, line 2: Replace "table 9.6" by
"table 9.7".
Page 250: The last sentence of the top paragraph contains
a typo: -2 should be 2, and Z(à,x) was not defined. For better
clarity, the last two sentences in that paragraph should read: "In
practice, this means that one evaluates Z(,x) at each map position,
x, where Z(à,x) is analogous to (9.9) with é1 replaced by x, and
is determined by the maximum of Z(à,x) at the given x value. Only
those points x are then excluded for which Z(,x)<2 and Z(x)<-2,
where Z(x) is the lod score under homogeneity."
Page 268, Solution 9.2, line 2: Replace "table 9.6" by
"table 9.7".
Page 270, line 1: Replace 1/3 by 2/3. Line 3: Replace
"with that mutation" by "without that mutation".
Page 279, ref. Hall et al. (1990): Replace "Anserson" by
"Anderson".
Page 294, line 2 up should read: "...tetraploid..."
Page 302, Support interval: Replace 110 by 55.
