2.4. SNPs Calling

Researchers selected SNPs when the MAF is more than 1% for human sequences, while they selected MAF > 5% for plant sequences. All of those are an estimate threshold. As we all know, different experiments may have their own errors and the sequence quality is also different when different technology platforms were used. In this study, we present a new way to find a reasonable MAF for each independent experiment. First we selected some stable genes which were already known as comparable samples and sequence with other samples together. Then the ratios of SNPs change by the MAF were calculated. To observe those trends of SNPs rations variation feature better, polynomial equation was applied to fit the curves (theoretically, N-order polynomial can approximate to any nonlinear function). We derived the first-order differential equation of fitting polynomial equation and that is the accelerating equation of initial equation. The stable value of the accelerated curve was the best threshold.

To check the result of SNPs' ratio by this process, the pretrimmed reads and original reads (clean and adapts discarded) were also used to map and screen SNPs. Three kinds of reads data were compared by SNPs' ratio and position. The assembled reads data should have less SNPs than other reads at the same MAF threshold.