IUBio

FASTLINK for DOS available

Alex Schaffer schaffer at cs.rice.edu
Thu Nov 17 20:09:13 EST 1994


The purpose of this note is to announce the availability of trial
versions of FASTLINK for DOS. Thanks to Ramana Idury and Roger Kou
at USC for figuring out how to convert the UNIX versions to
DOS versions.

FASTLINK contains significantly faster versions of the main programs
in LINKAGE.

Some fundamental similarities and differences between UNIX versions of FASTLINK
and DOS versions of FASTLINK should be noted:

Similarities:
1. DOS FASTLINK is essentially the same as FASTLINK 2.2, provided that
   the computer does not crash during the run.
2. Most of the documentation still applies except the instructions for
   installation (we distribute executables for DOS, so no compilation
   is required) and portability (basically irrelevant to DOS).
3. We are distributing versions of LODSCORE, ILINK, LINKMAP, MLINK, UNKNOWN.
   Other auxiliary programs can be obtained from the FTP site at Columbia.

Differences:
1. We are distributing only executable versions for DOS FASTLINK, rather
   than source as for UNIX FASTLINK. Details about how to get the executables
   are given below. We may distribute source at a later date.
2. DOS FASTLINK does not support crash recovery.
3. We are distributing only the "slow" versions of FASTLINK for now.
   These are still measurably faster than LINKAGE.

Cautions:
1. FASTLINK is derived from LINKAGE 5.1 for UNIX and not from LINKAGE for
DOS. We have tested FASTLINK on DOS, but we have not investigated to
what extent LINKAGE for UNIX and LINKAGE for DOS are incompatible.

2. If FASTLINK and LINKAGE give different answers, you should not
assume that FASTLINK is wrong. LINKAGE may be wrong.
I have fixed 2 significant bugs in LINKAGE LODSCORE/ILINK and 1
significant bug in LINKAGE LINKMAP/MLINK. These bug fixes are described
in README.updates. There is also an inconsistency in scaling in
LINKAGE LINKMAP, which is fixed in FASTLINK LINKMAP (see README.scaling).

3. You must a have 386 or higher machine. The code was tested on two
different 486 machines. I am fairly certain that it won't work on
286 and below machines.

4. You should have a Math co-processor on the machine.
Most recent 486 machines will have this.

5. We strongly recommend trying the programs outside of Windows
(i.e., in vanilla DOS) first. They probably work from Windows too,
but it may be the case that the compiler makes some assumptions
about how the video/monitor environment is set up, which are
not consistent with the setup of your machine.

Like FASTLINK 2.0, 2.1, and 2.2, DOS FASTLINK is
distributed by FTP from a computer at Rice University. 
I have not yet sorted out logistics for mailing out floppy disks.

Here are the instructions for retrieving the code:
   ftp softlib.cs.rice.edu

Login as anonymous and leave your full e-mail address as password.
   cd pub/fastlink/dos
   binary

In that directory you will find various files. You may wish to
retrieve all the files with the name README and all the files that
end .ps. These are all relevant documentation files.
The file README with no extension gives a roadmap to
all FASTLINK documentation.

Having retrieved whatever documentation you want you should retrieve:

   lo48.exe (LODSCORE with maxhap set to 48)
   lo96.exe (LODSCORE with maxhap set to 96)
   il48.exe (ILINK with maxhap set to 48)
   il96.exe (ILINK with maxhap set to 96)
   li48.exe (LINKMAP with maxhap set to 48)
   li96.exe (LINKMAP with maxhap set to 96)
   ml48.exe (MLINK with maxhap set to 48)
   ml96.exe (MLINK with maxhap set to 96)
   unknown.exe (UNKNOWN)

maxhap is the maximum allowed product of number of alleles at all loci
in the analysis. For example, if you want to do a 3-point analysis
with allele numbers 2, 3, and 3, the allele product is 18 and it is safe to
use the version with maxhap set to 48.
If you wish to do a 4 point analysis with allele numbers 2, 3, 3, and 4,
the allele product is 72 and you should use the version with
maxhap set to 96.
You should retrieve all 9 files. Remember to use binary mode
in FTP. When you wish to run the programs decide which version you want
to use and rename that to one of (lodscore.exe,ilink.exe,linkmap.exe,
mlink.exe).
Example:
   rename li96.exe linkmap.exe

Note: the limitation of maxhap set to 96 is experimental, not inherent.
The LINKAGE for DOS version distributed from Columbia use maxhap set to 48,
which is why we chose that as the lower value. While testing out the
code, we successfully tried a much larger value of maxhap (180) on a DOS
machine with 16Mbytes of RAM.

The decisions on how to set the constants in the executables we distribute
will undoubtedly be revisited often based on user comments.

Note: The DOS executable are not currently included in fastlink.tar.Z,
so you must retrieve them as individual files.

I would appreciate your feedback as to whether our DOS versions work
on your machine.

Our initial algorithmic improvements are described in the paper:

  R. W. Cottingham Jr., R. M. Idury, A. A. Schaffer. Faster Sequential
  Genetic Linkage Computations, American Journal of Human Genetics 53(1993),
  pp. 252-263.

The algorithmic improvements from version 1.0 to version 2.0
of FASTLINK are described in the papers:

  A. A. Schaffer, S. K. Gupta, K. Shriram, and R. W. Cottingham Jr.
  Avoiding Recomputation in Linkage Analysis, Human Heredity, 44(1994),
  pp. 225-237.

We ask you to cite both papers if you use FASTLINK in any
published experiments. Both papers are included in the distribution.

The code in this distribution is the result of complicated
collaborative work in which all 5 authors of the above papers played
significant roles.

The most important changes from FASTLINK 2.0 to 2.2 are described in
the file README.updates that comes with FASTLINK.

FASTLINK is a modified version of LINKAGE. You should continue to
cite the original papers on LINKAGE, if you use FASTLINK in a published 
experiment.

I am maintaining a e-mail mailing list of FASTLINK users. If you have
retrieved the code and would like to be on the mailing list, send me
e-mail at the address below. Being on the mailing list
is important to find out about corrected and improved versions of FASTLINK

Alejandro Schaffer, PhD
Department of Computer Science
Rice University
Houston, TX 77251
schaffer at cs.rice.edu
Tele: (713) 527-8101 x3813 (I do not have a direct line)
FAX:  (713) 285-5930








More information about the Bio-soft mailing list

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