Computer software to assist in identifying SNPS with microarrays
The present invention is a method to assist in the identification of single nucleotide polymorphisms (SNP) from microarray hybridization data. Data from hybridization protocols run on microarrays often have variations in the data resulting from variations in hybridization conditions and efficiencies and variations in optical intensities. An algorithm is described to screen the results to identify those data points most likely to be real SNPs as opposed to variations in the hybridization or sensing data.
1. A method for identifying in a sample single nucleotide polymorphisms relative to a reference nucleotide at a position in a reference sequence, the method comprising the steps of:
hybridizing the sample to a microarray of oligonucleotide probes, the microarray comprising, for each reference nucleotide, a first probe complementary to the reference sequence and second, third and fourth probes complementary to the reference sequence except for a non-matching nucleotide at the position, the four probes defining a position group at a physical space on the microarray, the nucleotide at the position on each probe in the position group being unique and being selected from the group consisting of A, G, C and T;
determining the relative hybridization intensity of each probe in each position group;
classifying each position group as a conformer if the highest intensity probe conforms to the reference sequence at the position, as a nonconformer if the highest intensity probe does not conform to the reference sequence at the position;
mapping the conformers and nonconformers to their physical space on the microarray;
identifying as possible SNPs only those nonconformers physically located adjacent on the microarray to conformers; and
outputting the identification of possible SNPs to a user.
2. A method as claimed in claim 1 wherein said mapping comprises using a machine learning algorithm to map the conformers and nonconformers.
3. A method as claimed in claim 2 wherein the machine learning algorithm is a nearest neighbor algorithm.
4. A method as claimed in claim 1 wherein the position is at a central position of each probe in the position group.