casava 1.8 bam conversion to gatk bam

[kingsalex]

Am am trying to convert bam files generated via casava to bam files recognized by gatk...I have asked illumina but no solution yet. I am using casava 1.8 to align and generate bam files(one bam file for all chr's). Illumina gave me a script which was supposed to work for conversion, but as far as I can see produces the same file as that when using a combination of picard reordersam + picard addorreplace, to add read groups and get correct contig order . This seems to have succeeded in terms of contig order and adding read groups, however, using GATK (DepthOfCoverage) on this 'converted' bam file I now get the error message:##### ERROR MESSAGE: Badly formed genome loc: Contig 1_gl000191_random given as location, but this contig isn't present in the Fasta sequence dictionary.

I have used the same ref fasta file in casava alignment as I have pointed gatk to (and picard when relevant), thus I don't understand why these additional/incompatible contigs are appearing. Perhaps workflow is dependant on a particualr reference file for alignment other than the one I am using (I am using human_g1k_v37.fasta reccommeded by gatk)?????

Any help would be greatly appreciated

[pzcity]

Have you generated the dict file from your fasta? Make sure the fasta does not have empty lines between the chromosomes.

http://picard.sourceforge.net/command-line-overview.shtml#CreateSequenceDictionary