The main purpose of CRI-MAP is to allow rapid, largely automated construction of multilocus linkage maps (and facilitate the attendant tasks of assessing support relative to alternative locus orders, generating LOD tables, and detecting data errors). Although originally designed to handle codominant loci (e.g. RFLPs) scored on pedigrees "without missing individuals", such as CEPH or nuclear families, it can now (with some caveats described below) be used on general pedigrees, and some disease loci.
This version of CRI-MAP is distributed as source code in the language C, on a 360K DOS formatted diskette. To use it, you will need to obtain access to a C compiler for your computer, and compile the program yourself (see the section GETTING STARTED, below). The present version of the program adheres fairly rigidly to the conventions of "standard C" as described in [ Kernighan and Ritchie (1978) ] (the main exception being that some variable and function names have more than 8 significant characters), and should be compatible with most C compilers.
The program was developed and tested using vers. 2.0 and 2.3 of the VAX C compiler, on a Microvax II with 5 Mb memory under the MicroVMS operating system. It has been successfully ported to a number of other computers, including other VAXes, SUN and Apollo workstations, and the Mac II. CRI-MAP requires a lot of memory; it is desirable to run it on a computer with at least 1 Mb (RAM or virtual memory) if you will be analyzing more than 10 loci simultaneously. It may be possible to run it on an IBM AT under DOS if your data set is small and you reduce the default memory allocations, although we have not tried this yet. A small data set with several chromosome 7 markers is provided with the program for the purpose of testing it (only). I would appreciate being informed of any difficulties in implementing the program, bugs, errors or gaps in the documentation, or suggestions for improvement.
Historical note: Version 1 of CRI-MAP was originally written by me in the summer of 1986 in the language APL; the portion of that version which does maximum likelihood estimation for a fixed locus order was based on algorithms developed in collaboration with Eric Lander [ Lander and Green (1987) ]. Collaborative's chromosome 7 map [ Barker et al., (1987) ] was constructed using that version of CRI-MAP, running on an IBM XT.
In the summer of 1987 parts of the original version were translated into C, with the help of Steve Crooks, and used in constructing the genome map published in Cell (Eric and his group at MIT independently constructed maps using the program MAPMAKER). At this time I also discovered the "layered EM" maximum likelihood search method, described in [ Green (1988) ].
In January, 1988, I worked out a much faster algorithm for likelihood calculation (the method of switch algebras -- actually a family of related algorithms), also described in [ Green (1988) ] . I have since written and tested the code for these new algorithms and incorporated them into version 2.0 of the program; Kathy Falls has worked on improving various other parts of the program. Versions subsequent to 2.3 were developed by me at Washington University.
I soon hope to incorporate a full likelihood analysis for pedigrees with missing individuals, and allow for disease loci with incomplete penetrance. Any user feedback will be gratefully received, and I will do my best to incorporate suggestions into future versions of the program.
Phil Green
Dept. of Genetics
Box 8232
Washington University School of Medicine
St. Louis, MO 63110
(314) 362-5192
Fax: (314) 362-4137
email: pg@genome.wustl.edu
[Note by wli(9/31/95): this address from 1990 is no longer correct in 1995. The current email address is phg@u.washington.edu.]
next section: 2. using cri-map with general pedigrees and disease loci