I downloaded two BAM files from the TCGA Legacy repository, a normal BAM and a tumor BAM for the same patient. I tried to process the files with Mutect2, but I got an error:

<h5>ERROR AN INPUT FILE ERROR has occurred (version 3.7-0-gcfedb67):</h5> <h5>ERROR MESSAGE: Cannot merge sequence dictionaries because sequence X and 22 are in different orders in two input sequence dictionaries.</h5>

I regenerated the BAM indexes with samtools index and checked the BAMs with picard's validation tool, I got this error.

HISTOGRAM java.lang.String

Error Type Count

ERROR:CIGAR_MAPS_OFF_REFERENCE 16

ERROR:INVALID_INDEXING_BIN 6961

I tried to process the files with hg19 and GRCh38.d1.vd1 reference genomes.

What is the problem and how can I solve it?