First base of reads SNP or false result?

[pirates.genome]

Hi,
I am using 5 samples pooled DNA Targeted sequencing data (Illumina FASTQ) and using BWA - GATK for mapping and variant calling. In variant calls (vcf files), I find that at aparticular position of genome, first base of the 900 reads (coverage is 1000X) showing mutation (889 out of 900 were showing altered base) but GATK gives very low QUAL score ~40, while highest QUAL in dataset is something about 20,000.

My question is first base is of high quality base calling and so high altered base count makes it good candidate of mutation. Why does GATK gives so much low QUAL score?

[swbarnes2]

I think the first base is mroe likely to be an error than bases a little further in. Or, you have a PCR mistake that became compounded with PCR duplication. Or those reads are misaligned, and really belong somewhere else.

A real SNP should show up in a lot of reads that start at a lot of places.

[pirates.genome]

Thank you swbarnes2 for explaining why only first base is not so good as variant rather true SNP should be covered by many different reads starting at different positions.

It is targeted sequencing, one of our targeted exon has 900X coverage and mutational hotspot within that exon was our target. Saying in other ways, when we target a very specific region, amplify using PCR and map them back on genome, obviously they show a similar kind of sequence reads, most of them starting at the same position.

Coming to the question, in such type of study, should we believe a variant in the very first base of all reads or not?