I was given some bam files I used to call somatic variants. The high number of SNP and mutations i found seemed suspicious, so I asked for the original fastq files to redo the allignment. While waiting for them I tried to get fastq from those bam files using bedtools bamtofastq . I used bwa mem to check allignment and the new bam file yeld less mutations and SNPs. Is there a procedure I didn't understand in the bedtools function that might have given me a fastq file different than the original?