FINGCOMP 04.12.95

The program FINGCOMP allows calculation of the joint likelihood
of multilocus DNA fingerprints observed in a pedigree of given 
structure. FINGCOMP requests an input file of the following
format:

1. line; NPOS, a single positive integer indicating the total 
number of positions to be considered for each fingerprint. If 
a grid is assumed to be superimposed upon the autoradiograph(s) 
so as to digitalise the profile(s) (+ band present, - no band 
present), then NPOS is the total number of positions defined 
by this grid.

Following lines; for each individual enter 
ID         a positive integer to identify that individual,
PID1 PID2  the identifiers of the parents (enter 0, if a parent
           is not present, but remember that you might have to 
           include dummys in order to define the correct pedigree 
           structure), 
NBAND      the number of bands observed for that individual (enter
           -1 if a profile is not available),
POS(1) ... 
POS(NBAND) the position(s) at which a band has been observed, i.e.
           NBAND positive integers.

All entries have to be separated by at least one blank. The data
can be distributed over any number of lines. Following the information
for the last individual in your pedigree, enter -1 to indicate the
end of the data file.

6
 10   0   0  3  1 2 4
  5   0   0  3  1 3 6 
 19   0  10 -1
234   5  19  3  1 2 3
-1

This sample file would correspond to the following pedigree and profiles:

H1 (hypothesis 1)
    
  10            10   5  234
   ³             +   +   +
  19ÄÂÄ5         +   -   +
     ³           -   +   +
    234          +   -   -
                 -   -   -
                 -   +   -

For a bandsharing of 0.25, these data yield log10(likelihood) = -5.83277.

If we wish to test whether 19 is the father of 234, we have to compare
the above log10(likelihood) with that obtained under the following 
hypothesis:

H2 (hypothesis 2)

  10            10   5  234
                 +   +   +
     ÚÄ5         +   -   +
     ³           -   +   +
    234          +   -   -
                 -   -   -
                 -   +   -

The corresponding file reads

6
 10   0   0  3  1 2 4
  5   0   0  3  1 3 6 
234   5   0  3  1 2 3
-1

Note that individual 10 has to be included since this is critical 
for the likelihoods to be comparable. In general, all individuals 
for which phenotypic information has been considered under one 
hypothesis have to be included under any other hypothesis to be 
compared with. 

The above data yield log10(likelihood) = -6.201186, which implies that 
the log10-likelihood ratio is -5.83277 - (-6.201186) = 0.36842 in favor
of H1. In other words, the fingerprints are 10^0.3682 = 2.3 times
more likely under H1 (individual 19 is the father of 234) than under H2 
(individual 10, and therefor individual 19, is not related to 234).

Best wishes
MK