Relationship between depth and genotype?

Hi, I am new to WGS and would like to know how the genotype is attributed. It seems it does not depend much on depth of allele ratio. We ran GATK3.6 and Freebayes on the same BAM 50-50X files and intersected them to reduce the number of variants between monozygotic twins (300k per caller). We got 16k residual variants after intersect that we need to filter and prioritize. GATK has much better concordance of allele ratio with genotype than Freebayes which gives much higher coverage. (1) What is the relationship between coverage (individual allele depth) and genotype? (2) Should we trust more allele balance or depth upon filtering? (2) What is the best practice to filter out variants from WGS? Thank you -Anne