A serious potential pitfall in the LINKAGE programs was recently uncovered by one of my collaborators, Dr. Chantal M‚rette. She analyzed a pedigree for linkage between a dominant disease and a marker with 8 alleles. The data clearly showed a recombination, but the MLINK program (version 5.04) did not report a lod score of minus infinity at é=0. The explanation was as follows. The marker was coded as an allele numbers locus type with 8 alleles. In LINKAGEC.PAS, there is a program constant, MAXFACT, which gives the maximum number of binary codes at a single locus. MAXFACT was left at its former value of 5 since no binary factors loci were to be analyzed. As it turns out, MAXFACT is also relevant for allele numbers loci and must be at least as large as the number of alleles, MAXALL. In future versions, a check will be incorporated to verify that this is indeed the case. For now, users should verify manually that MAXFACTòMAXALL in LINKAGEC.PAS.
Conversion of the LINKAGE programs to the Macintosh (Dr. D. Weeks) is still in progress but has been relatively slow due to time constraints. These programs should become available towards the end of 1990. They will be announced in this newsletter. It should be noted, however, that presently there are no plans to also have LCP (Linkage Control Program) running on the Macintosh, since converting LCP by us would be too much of a time commitment.
Adapting the LINKAGE programs to OS/2 is going smoothly. The problems with the Prospero Pascal compiler reported in the previous issue of this newsletter have been resolved. The current version of the compiler, version iio 5.202, is free from the bugs seen in previous versions. It is available from Prospero Software, 100 Commercial Street Suite 306, Portland, Maine 04101, and is highly recommended. The purchase price ($253 after educational discount, plus $5 handling charge) includes both the MS-DOS and OS/2 versions. It has several advantages over Turbo Pascal. For example, programs need not be broken into smaller units for compilation; file buffers can be set for any files, not just text files as in Turbo Pascal; up to 4 MB of memory can be addressed dynamically under OS/2 so that larger problems can be run (Turbo Pascal is not available under OS/2). ProPascal shares with Turbo Pascal the restriction that no Pascal procedure can be larger than 64KB in code (none of the procedures in the LINKAGE programs exceeds that limit). On the other hand, our test problem runs 34% faster when compiled with Turbo Pascal than with ProPascal (for DOS). We may be able to increase speed under ProPascal somewhat but probably not up to the level of Turbo Pascal. More details will be given in the next issue of the newsletter.
Several of the HOMOG programs have been updated. The HOMOG2 program has been changed to report likelihood ratios instead of p-values for the test of homogeneity versus heterogeneity. Since even the asymptotic distribution of the test statistic is unknown for the case addressed by the HOMOG2 program (when going from two to one component, this single restriction automatically eliminates two parameters), the formal p-values seem unreliable. Also, a specialized version of the HOMOG3 program has been added: HOMOG3R assumes that a locus is linked to either of two markers on different chromosomes, or is unlinked with the two markers. As usual, new program versions are obtained by requesting from us a list of programs with ordering instructions.
Jürg Ott August 1990
Columbia University, Box 58 FAX +1 (212) 568-2750
722 West 168 Street Tel. (212) 960-2507
New York, NY 10032 Bitnet: OTT@NYSPI
Advanced Linkage Course
Thursday through Saturday, November 8-10, 1990
Course Description and Prerequisites
This course is intended for researchers who already have
experience with linkage analysis and would like to become more
proficient in analyzing linkage data and handling unusual problems.
A working knowledge of IBM PC's is assumed. The course will begin
with a brief introduction to theory and techniques. Its main part
will consist of problem-solving sessions with use of computer
programs (LINKAGE version 5.04 or, hopefully, version 5.1; MENDEL;
HOMOG; Linkage Utility Programs) and general question-and-answer
sessions. The topics to be covered comprise heterogeneity,
pseudoautosomal linkage, inbreeding loops, risk calculations,
sex-specific recombination fractions, etc. Computer programs can
be copied (on 3«" or 5¬" diskettes) and taken home. A manual will
be mailed before the start of the course.
The course will be taught by myself and my collaborators. It
will take place in the classroom of the Health Sciences Library
(701 West 168th Street) which is equipped with 20 microcomputers of
type IBM PS/2 (3«" diskette drives). Due to space limitations,
course attendance is limited to 30 participants. Participants
should plan on arriving in New York in the evening of Wednesday,
November 7, 1990.
Course Fee
The fee for the 3-day-course is $400 for researchers at an
academic institution, and $500 for individuals from private (for
profit) companies. It may be paid by a check drawn on a U.S. bank,
made payable to Columbia University Dept. of Psychiatry, by
Government pay order, or by Travellers checks, but send no money
now -- applicants will receive a bill and information regarding
cancellation policy. As there is presently no support for this
course from sources other than the course fee, no reduction of the
cost to applicants is possible. This fee covers tuition and course
related expenses (handouts, diskettes, etc.) but not room, board or
meals. Course participants will receive a list of good and
moderately priced hotels in New York and will have to make their
own arrangements (except foreign participants). A small number of
guest rooms are available in Bard Hall next to the Health Sciences
Library (double rooms at $60 per room per night, single rooms at
$55).
Application for Linkage Course by Dr. Jürg Ott
Please fill out this page and send it by mail or, preferably,
by FAX. People interested
in staying in a double room at Bard Hall should indicate a
preferred roommate, or else we will match applicants.
Your name: _______________________________________________
Affiliation:_______________________________________________
Address: _______________________________________________
_______________________________________________
_______________________________________________
Tel. number:_______________________________________________
FAX number: ______________________ Bitnet:_________________
Interested in staying at Bard Hall? YES / NO DOUBLE / SINGLE
If yes, which nights? Wed Thu Fri Sat
What size diskettes do you use? 3«" 3«"HD 5¬"HD (please circle)
(we prefer 3«"; we have no easy way of writing to low-density 5¬"
disks)
For applicants from abroad: Do you want us to make hotel
reservations? What accommoda- tions?
Below, please describe which linkage programs you have used if any,
how many families you have analyzed, and other experience in
linkage analysis you might have:
Signature: ____________________________Date: ___________