Other than the reference sequence, I'm assuming that the most significant edit that I have to make to the standard protocol, is changing the genome sizes (from human to fly) in the parameters file.
The most recent assembly for D.melanogaster (dm6) has 1863 unmapped contigs/scaffolds, ranging in size from 1000bp to 20Kb.
Some of these unmapped scaffolds are from chrY (e.g chrY_DS485315v1_random) , and some are from chrX (e.g. chrX_DS485798v1_random).
My current plan is to use only the seven useful chromosome scaffolds, (using -L in the preparation stage), and present the genome sizes of male and female in the parameters, calculated from the lengths of useful chromosomes only.
Does this seem like a reasonable plan?
An alternative would be to add-on the lengths of all unmapped chrX scaffolds to the length of chrX, and the same for chrY, when calculating the different male and female haploid genome sizes.
(I just noticed a reply to yesterday's related question http://gatkforums.broadinstitute.org/discussion/comment/26011#Comment_26011 )