Q1. 53051 in total, 48687 that pass Q30.
Q2. This is a hard question, we can see that at Truth:90 we are only getting around 15K snps, but all are predicted to be true. If we increase to Truth:99 we 41k SNPs, but when we begin to get some False Positives (orange). So selecting a threshold is a tradeoff between specificity and sensitivity. Lets use the 99.0 threshold, but maybe we should. 
Q3. SNPs that pass: 44298 (99.0)
Q4. I would not trust Novel SNPs with low quality score
Q5. If most Asian SNPs are already in dbSNP then "novel" ones that we find are likely to be not true.
Q6. 43895, ie we only removed 44298-43895=403 novel SNPs
Q7. Hard filtering: 37932 ; Soft+Min5DP: 40279 ; overlap: 35052
Q8. Heterozygotes: 25094, Homozygotes: 18783
Q9. HC calls: 49338, overlap: 48561, UG: 53051 (question 1). Probably the UG has many False Positives (the non-overlapping).
Q10. samtools: 43381, gatk: 43895, overlap: 38277