System, method, and computer software product for genotype determination using probe array data

1. A method for determining the genotype of a plurality of nucleic acid samples at a plurality of SNPs, comprising:

(a) hybridizing each nucleic acid sample in said plurality of nucleic acid samples to an array of allele specific probes to obtain a plurality of raw probe intensity measurements;

(b) normalizing said raw probe intensity measurements to obtain normalized probe intensities;

(c) summarizing the normalized probe intensities to obtain an allele signal estimate for each allele of each SNP in each nucleic acid sample, wherein the allele signal estimate for each allele of each SNP in each nucleic acid sample comprises a first value S A and a second value S B where S A is the summary value for a first allele of the SNP and S B is the summary value for a second allele of the SNP;

(d) transforming said allele signal estimates in clustering space in one-dimension to obtain a transformed allele signal estimate for each SNP in each nucleic acid sample using the following equation: transformed allele signal estimate =asinh(K(S A −S B )/(S A +S B ))/asinh(K), where K is a tuning constant;

(e) obtaining a prior distribution of genotype cluster characteristics;

(f) evaluating all possible assignments of the transformed allele signal estimates to the genotype clusters of the prior distribution;

(g) calculating an optimal assignment from the possible assignments of step (f) for each transformed allele signal estimate, to one or more genotype clusters using a Gaussian cluster model;

(h) updating the prior distribution with optimal assignments calculated in (g) to obtain a posterior distribution of genotype cluster characteristics for each SNP; and

(i) using the posterior distribution of genotype cluster characteristics for each SNP to make genotype calls for that SNP in each nucleic acid sample.

2. The method of claim 1 wherein the array comprises more than 1 million different sequence probes and wherein each probe is perfectly complementary to an allele of a SNP that has a minor allele frequency of at least 2% in a population.

3. The method of claim 1 further comprising obtaining a confidence value for each of the genotype calls.

4. The method of claim 1 wherein the step of normalizing said raw probe intensity measurements comprises a quantile normalization step, a log-scale transformation and a median polish step.

5. The method of claim 1 wherein K is 2.

6. The method of claim 1 wherein K is 1 to 4.

7. The method of claim 1 wherein the prior distribution of genotype cluster characteristics is a generic prior derived from a plurality of different SNPs.

8. The method of claim 1 wherein the prior distribution of genotype cluster characteristics is a specific prior for each SNP that has been derived using data for that SNP in a plurality of samples.

9. The method of claim 1 wherein there are three possible genotype clusters including in the prior distribution: a first cluster of samples that are homozygous for the first allele, a second cluster of samples that are homozygous for the second allele and a third cluster of samples that are heterozygous and wherein the third cluster is between the first cluster and the second cluster.

10. The method of claim 1 wherein the posterior distribution of genotype cluster characteristics for each SNP comprises a cluster center for each of three possible genotypes, AA, AB and BB, and covariance matrices for each of the three possible genotypes.

11. The method of claim 10 wherein the genotype call is made by obtaining relative probabilities p(AA), p(AB) and p(BB) from the cluster centers and covariance matrices and calling the genotype with the highest probability for each SNP.

12. The method of claim 10 wherein an ordered relationship between the centers of the genotype clusters is required during calculating step (g).