Hi!
We are working on various non-model organisms without snp databases available. Of course this makes BQSR a problem, and we're not entirely convinced by the "bootstrapping" method.
However the question came up if alternative base callers can't solve this problem at least partly? There are of course many, such as Ibis, AYB and BayesCall, that can be used as an alternative to Bustard. Some of these programs supposedly have a better quality score distribution [1]. Have these ever been tested with GATK?
1 Ledergerber, Christian, and Christophe Dessimoz. 2011. “Base-Calling for next-Generation Sequencing Platforms.” Briefings in Bioinformatics 12 (5): 489–97. doi:10.1093/bib/bbq077.