IUBio Biosequences .. Software .. Molbio soft .. Network News .. FTP

A decent program for Linkage Analysis

Gerard Tromp tromp at sanger.bcm.tju.edu
Sat Apr 29 12:28:04 EST 1995

In article <D7rHDL.BJ5 at news.cis.umn.edu>, dean at birch.med.umn.edu (Dean Flanders) writes:
> : can anyone tell me where I can get a straightforward linkage
> : program with good documentation.  I have looked at Linkage 5.1
> : and Mapmaker and these do not fit either criteria
> Now this is a decent question.  You are right there is a learning curve 
> to these programs, however, you will experience that with any linkage 
> analysis program you run.  Sorry, no warm fuzzy interfaces on any 
> linkage analysis program I have ever run!  I would recommend taking Jurg 
> Ott's linkage course or getting Joseph Terwilliger's book Handbook of 
> Human Genetic Linkage.
> : obscure, badly written programs, with wretched documentation.
> : We have been using the original version of Linkage, which is
> : adequate, but not up to snuff. I wonder if someone has a version
> : of linkage between the 5.1 and 0.9 version (which we have).
> Now this is a little irritating.  I think in the linkage community that 
> the consensus is that these are extremely well-written programs, and just 
> because you do not understand how to run them does not automatically make 
> them "obscure, badly written programs, with wretched documentation".  Not 
> to mention that people are still re-writing and improving on the 
> original versions.  Yes, these programs are complicated, but so is linkage 
> analysis.  You have just exhibited your naivete in this area to me and 
> countless others.
> I will send you a copy of the bionet.molbio.gene-linkage FAQ, which 
> lists some other linkage programs.  I must admit you may find them 
> "obscure, badly written programs, with wretched documentation", 
> because every single one of them has a learning curve.  However, you  
> could write your own programs, of course.
> : Thanks for the help. Any suggestions would be greatly
> : appreciated.
> You're welcome.  One other suggestion, think before you post!  If you think 
> I am irritated, I would hate to hear what the people who wrote these 
> programs think of your post.
> Have a nice day,
> Dean
	I wish to second Dean Flanders' reply. The learning
curve for linkage is steep, particularly for those of us who do
not have a formal background in statistics and likelihood
theory. The programs are based on paramtric linkage analysis one 
has to know how to parameterize the linkage model/problem. 
The programs are confusing in the beginning because one needs to 
know a) how to parametrize the problem and b) how to present it
to the programs. The book by Terwilliger and Ott, "Handbook of Human
Genetic Linkage" is lucid with many worked-out examples. As such
it is the documentation for the LINKAGE programs which makes
them rather well documented. It is true that the original
documentation accompanying the LINKAGE programs was somewhat
cryptic, however, they were meant as reminders for people who
had attended the linkage workshop/course that Dean mentions.
	The statement that the programs are poorly written is
inaccurate. Yes, almost all programs can be improved, and some
unneccssary recalculation and recursion were removed form the
linkage code by Alex Schaffer and the group at Rice when they
generated FASTLINK. However, if one looks at the original code,
it is definitely not poorly written. Some bugs have been found
and corrected, but a few bugs in a program do not constitute bad
programming, particularly when the problems are complex and it
is well nigh impossible to test the programs under all
permutations of possible problems.

Gerard Tromp, Ph.D.
Research Assistant Professor		Vox:	215-955-4487
Department of Biochemistry			215-955-4488 
	and Molecular Biology		Fax:	215-955-5393
Thomas Jefferson University
233 South Tenth Street, Room 328	E-mail:	tromp at sanger.bcm.tju.edu
Philadelphia, PA 19107 

More information about the Gen-link mailing list

Send comments to us at biosci-help [At] net.bio.net