
PREST (Pedigree Relationship Statistical Test) Changelog


Version 1.0 (7/22/00)
=======================
- Initial Release


Version 1.01 (7/25/00)
=======================
- Changed program behavior so that it does not halt after
  encountering a Mendelian error.  Instead, just warn user of the
  error.


Version 2.0 (8/10/00)
=======================
- Extend the the set of alternative relationships for the MLRT.
  Version 1 considers full-sib, half-sib, grandparent-grandchild,
  avuncular and first-cousin (excluding the null relationship).
  Version 2 considers all the above relationships and three more
  relationships. They are half-avuncular, half-cousin and 
  half-sib+first-cousin.  (A pair is called half-sib+first-cousin
  if they have the same mother and different fathers and the two fathers
  are full-sib, or the same father and different mothers who are sisters.)

- Provides a new program, ALTERTEST, which performs the EIBD, AIBS, IBS
  and MLLR tests on relative pairs with a user-specified null hypothesis
  that may be different from what the pedigree says. 

  
Version 2.01 (9/05/00)
=======================
- Changed one of the input files "altertest_input" for ALTERTEST 
  so that the input file does not require the information of the 
  putative relationship type according to the pedigree of the pairs.

- Changed ALTERTEST behavior so that it also estimates p = (p0, p1, p2),
  the probabilities of sharing zero, one or two alleles IBD, for each 
  of the pairs.


Version 2.02 (10/30/00)
=======================
- Some changes in the output file "prest_out1" for PREST.

- Some changes in the documentation.


Version 2.03 (11/28/00)
=======================
- Increased the default maximum line length (set in compiler variable
  MAXSTR) of the input files to 5000.  The program now warns if an input
  line is too long.

- The PREST distribution now includes an R program written by
  Daniel E. Weeks that draws out the relationship triangles using
  the PREST output file "prest_out2".  That is, it will plot the sharing
  probabilities on the 'relationship triangle' as described in
  Elizabeth Thompson's book "Pedigree Analysis in Human Genetics".
  This program runs fine in Splus as well as R.  We added a minor
  modification so that it works with "prest_out2" files created using
  either option of PREST.


Version 3.0 (01/10/01)
=======================
- Add 2 additional relationships to the alternative set A for the 
  MLLR test.  They are parent-offspring and MZ twin relative pairs.

- Add a new parameter 'errorrate' (the approximate genotyping error 
  rate, the default value is 0.01).  Incorporate genotyping error in the
  likelihood calculation for the parent-offspring and MZ twin pairs. 

- Change PREST behavior so that, among typed individuals of each pedigree,
  it also identifies relative pairs that are half-avuncular, 
  half-first-cousin, half-sib-plus-first-cousin or parent-offspring
  relationship specified by the pedigree.  PREST then performs the 
  corresponding hypothesis tests.  

- Add new initial screening criteria for the PREST:
	1. for parent-offspring pairs, apply the MLLR test if the 
	   estimate of p1 < 0.9.
	2. for non-parent-offspring pairs, in addition to the criteria 
           used in Version 1 and 2, apply the MLLR test if the 
           estimate of p1 > 0.75.

- Add a new output file "prest_out3" for PREST. This file
  contains the information for the pairs that the pedigree indicates 
  not parent-offspring relationship but have the estimate of p1 > 0.75,
  and it also contains the information for the pairs that the pedigree
  indicates parent-offspring relationship but have the estimate of 
  p1 < 0.9.
 
- Change ALTERTEST behavior so that it also allows the 
  parent-offspring and MZ twin relationships as null hypotheses 
  for all the tests.


Version 3.01 (04/30/03)
=======================
- Correct the kinship coefficient for half-first-cousins to be .03125 
  in prest.c, altertest.c and tips.txt

- Add the feature of checking allele frequency to be >0 and <=1.
  (Thanks to David Poznik who identified the problem in the EM algorithm,
   due to ultra-low frequency of some alleles).
  
- In the EM algorithm for estimating p0, p1, p2, modifying the conditional 
  probabilities, P(genotype|IBD=i), for parent-offspring pairs, in a similar
  fashion as for the likelihood calculation, so that the estimates are robust
  to the presence of genotyping error.


Version 3.02 (02/15/05)
=======================

- There were a couple lines in the .c files that had illegal line-breaks,
  which either gives warning of "multi-line string literals are deprecated"
  during compiling or failure to compile. 
  (Thanks to David Poznik who identified the problem)

