2) Assignment or exclusion of individuals (or groups of individuals)

Start GeneClass2 from your "Start" menu, folder "CBGP". The splash-screen appears, then the main window of GeneClass2.

Load a reference datafile by clicking on the upper "Open" button [H1] and choose the datafile in the selector.
Load a to-be-assigned individuals datafile by clicking on the lower "Open" button [H1'] and choose the datafile in the selector.
Note that for self-assignment purposes only a single file is needed (i.e. the reference file [H1]) .

Choose "Assignment" [H2] in the first tab of the main window and "Individuals" [H3] in the "Assign:" box.
Note that assignment statistics can be also computed for groups of individuals rather than individual genotypes (Baudouin & Lebrun, 2000). Groups of individuals must be coded as populations in the samples file.

The assignment threshold (without probability computation) can be set by sliding the "Assignment threshold of scores" cursor [H4] (see computations without associated probabilities section for a definition of scores).



Click on the "2) Criteria for Computation" [I1] tab and choose the desired criterion (e.g. "Rannala & Mountain" [I2], see Rannala & Mountain, 1997 and Cornuet et al., 1999). Bayesian and Frequencies-based methods appears to be better than distance-based methods (see Cornuet et al. 1999 for a comparative study).


If you want to compute the probability that an individual belongs to each reference population, click on the "3) Probability computation" tab [J1], and then check the "Enable probability computation (Monte-Carlo resampling)" box [J2].
You can now choose a resampling algorithm [J3], eg. Paetkau et al. (2004) (recommended, but see Rannala & Mountain 1997 and Cornuet et al. 1999). Slide the cursor [J4] to define the minimum number of simulated individuals (eg. 1000, default value or 10000, leading to ten times longer but more precise computation), and the cursor [J5] setting the type one error eg. 0.01, default value (see Cornuet et al. 1999; Paetkau et al. 2004 ).


If needed, you can deselect some loci in the "Loci selection" tab [K1]. Deselected loci will be ignored during calculations.
Finally, click on the "Start" button [K2] to run the computation.


The program displays the "Log file" [L1] that displays the running parameters.
The progress bar and a counter shows the state of the computation [L2].
The "Stop" button [L3] allows the computation to be aborted.


Once computations are finished, the results are displayed in a grid in which the probability for each individual belonging to each reference population is displayed [M1] (Cornuet et al., 1999). If a given individual's probability in a given reference population is lower than the previously defined threshold, the value is greyed [M2].
Results can be printed ("Print" button [M3]) or exported in csv format ("Export" button [M4]).

 


Computations without associated probabilities:

If you did not check the "Enable probability computation (Monte-Carlo resampling)" box [J2] the results are displayed as follows: The five first pairs of columns show, for each individual, the most likely populations and their relatives scores (cf. note below) in decreasing order [N1]. The right part of the grid displays -log values of the likelihoods [N2].
The number of displayed scores can be changed with the spin edit area [N3].

Note: In a reference file with k populations the score of an individual i in a population l is computed as follow:

,

with Li,l the likelihood value of the individual i in the population l.


To main page