I have a population of 60 relatively closely related samples (non-human). After running through the GATK best practices to identify variants (with hard-filtering instead of VQSR), the phylogenies I reconstruct with the SNP VCF file have huge tips relative to internal branches, including the same sample that was sequenced twice. 99% of the variation between A and B is the same variation that exists between A and any other sample. For example, say Sample01A and Sample01B are differentiated by 1000 SNPs, while samples 01A and 60 are differentiated by 1050 SNPs, where 01A and 01B are two preps of the same sample.
Is this a symptom of a typical mis-use of GATKs pipeline? Or is this more likely occurring because of my samples specifically, such as my samples are too closely related and sequencing error is drowning out real variation?
Thanks for any suggestions, if possible,
Alex