Methods for identifying DNA copy number changes using hidden markov model based estimations
View Patent ↗Methods for estimating genomic copy number and loss of heterozygosity using Hidden Markov Model based estimation are disclosed.
1. A computer-implemented method for estimating a copy number of each of a plurality of genomic regions in a nucleic acid sample comprising a plurality of nucleic acid molecules, each genomic region containing at least one single nucleotide polymorphism (SNP), the method comprising:
hybridizing the plurality of nucleic acid molecules with a plurality of allele-specific perfect match probes for at least one SNP on a nucleic acid array comprising a plurality of perfect match probes wherein each of the plurality of perfect match probes does not have a corresponding mismatch probe;
obtaining, by a computer comprising a processor and a memory, an initial intensity measurement for each of the plurality of allele-specific perfect match probes wherein the initial intensity measurements obtained for the plurality of allele-specific perfect match probes for the SNP do not include intensities for any mismatch probes;
normalizing, by the processor, the initial intensity measurement for each of the plurality of allele-specific perfect match probes, resulting in normalized intensity measurements, wherein the normalized intensity measurements are determined for the plurality of allele-specific perfect match probes without utilizing data derived from any mismatch probes;
determining and applying a cross-talk correction without use of mismatch probe data, by performing a rotation transformation to obtain cross-talk corrected normalized intensity measurements of the plurality of allele-specific perfect match probes, wherein the cross-talk corrected normalized intensity measurements include a cross-talk component from another allele that is determined to correct for probes for one allele of an SNP hybridizing with another allele of the SNP;
calculating, by the processor, an initial copy number estimate for each of the plurality of genomic regions, wherein the initial copy number estimate is based upon the cross-talk corrected normalized intensity measurements and the cross-talk corrected normalized intensity measurements of a reference sample;
performing, by the processor, data smoothing on the initial copy number estimates, wherein the data smoothing reduces noise within the initial copy number estimates to generate smoothed copy number estimates; and
estimating, by the processor, the copy number of each of the plurality of genomic regions using a Hidden Markov Model to assign the smoothed copy number estimates to different copy number states.
2. The method of claim 1 , further comprising an allele-specific summarization step, wherein summarization values for each allele of the at least one SNP are generated based on the normalized intensity measurements.
3. The method of claim 1 , further comprising a filtering step, wherein the initial intensity measurements are filtered to include only intensity measurements from allele-specific perfect match probes corresponding to SNPs that satisfy one or more user-defined metrics, including SNPs on PCR fragments having a fragment length or GC content within a range of threshold values.
4. The method of claim 1 , further comprising a filtering step, wherein the initial intensity measurements are filtered to remove initial intensity measurements based upon one or more metrics, including at least chemical or scanner probe saturation.
5. The method of claim 1 , wherein performing data smoothing comprises a method of smoothing selected from the group consisting of Gaussian smoothing, Lowess smoothing, and Spline smoothing.
6. The method of claim 1 , wherein the plurality of allele-specific perfect match probes is located on two or more arrays, wherein the method further comprises a virtual array generation step, and wherein the virtual array reduces variance between the two or more arrays.
7. The method of claim 1 , wherein the normalizing step comprises median scaling, wherein a median value is computed based on intensity of all of the plurality of allele-specific perfect match probes.
8. The method of claim 1 , wherein the initial copy number estimates are calculated based on log 2 transformations of the normalized intensity measurements.
9. The method of claim 1 , wherein the initial copy number estimates are calculated using the following equation:
T
C
N
=
log
2
[
S
A
R
A
+
R
B
+
S
B
R
A
+
R
B
,
]
wherein S A comprises the normalized intensity measurement for a first allele of a SNP within the nucleic acid sample, S B comprises the normalized intensity measurement for a second allele of the SNP within the nucleic acid sample, R A comprises the normalized intensity measurement for the first allele of the SNP within the reference sample, and R B comprises the normalized intensity measurement for the second allele of the SNP within the reference sample.
10. The method of claim 1 , wherein the reference sample is a global reference sample generated from a plurality of control samples.
11. The method of claim 10 , wherein the plurality of control samples comprises between 10 and 40 samples.
12. The method of claim 11 , wherein the plurality of control samples is matched to the nucleic acid sample in either ethnicity or gender.
13. A system for estimating a copy number of each of a plurality of genomic regions in a nucleic acid sample comprising a plurality of nucleic acid molecules, each genomic region comprising at least one single nucleotide polymorphism (SNP), the system comprising:
a high-density oligonucleotide array comprising a plurality of perfect match probes wherein each of the plurality of perfect match probes does not have a corresponding mismatch probe, and the array is configured to hybridize at least some nucleic acid molecules in a nucleic acid sample with a plurality of allele-specific perfect match probes for at least one SNP on the high-density oligonucleotide array;
a computer; and
a computer software program configured to execute on the computer, causing the computer to:
obtain, from the high-density oligonucleotide array, an initial intensity measurement for each of a plurality of allele-specific perfect match probes for the at least one SNP in each genomic region, wherein the initial intensity measurements obtained do not include intensities for any mismatch probes;
normalize the initial intensity measurement for each of the plurality of allele-specific perfect match probes, resulting in normalized intensity measurements, wherein the normalized intensity measurements are determined for the plurality of allele-specific perfect match probes without utilizing any data derived from any mismatch probes;
determine and apply a cross-talk correction without use of mismatch probe data, by performing a rotation transformation to obtain cross-talk corrected normalized intensity measurements of the plurality of allele-specific perfect match probes, wherein the cross-talk corrected normalized intensity measurements include a cross-talk component from another allele that is determined to correct for probes for one allele of an SNP hybridizing with another allele of the SNP;
calculate an initial copy number estimate for each of the plurality of genomic regions, wherein the initial copy number estimate is based upon the normalized intensity measurements and a reference sample;
perform data smoothing on the initial copy number estimates, wherein the data smoothing reduces noise within the initial copy number estimates to generate smooth copy number estimates; and
estimate the copy number of each of the plurality of genomic regions using a Hidden Markov Model to assign the smoothed copy number estimates to different copy number states.
14. The system of claim 13 , wherein the computer software program is additionally configured to output to a display an indication of the copy number for each of the plurality of genomic regions.
15. The system of claim 13 , wherein the computer software program is additionally configured to output to a display an indication of an allele-specific copy number for each of the plurality of genomic regions.
16. The system of claim 13 , wherein the computer software program is further configured to filter the plurality of allele-specific perfect match probes to exclude probes based on at least one of a length and a GC content of a nucleic acid molecule on which the at least one SNP resides.
17. A computer-implemented method for estimating a copy number of each of a plurality of genomic regions in a nucleic acid sample comprising a plurality of nucleic acid molecules, each genomic region containing at least one single nucleotide polymorphism (SNP), the method comprising:
hybridizing at least some nucleic acid molecules in a nucleic acid sample with a plurality of allele-specific perfect match probes for at least one SNP provided in a high-density oligonucleotide array comprising a plurality of perfect match probes wherein each of the plurality of perfect match probes does not have a corresponding mismatch probe;
filtering the allele-specific perfect match probes for the SNPs based on one or more user-defined metrics;
obtaining, by a computer comprising a processor and a memory, an initial intensity measurement from the high-density oligonucleotide array for each of a plurality of allele-specific perfect match probes for at least one SNP, wherein the initial intensity measurements obtained do not include intensities for any mismatch probes;
normalizing, by the processor, the initial intensity measurement for each of the plurality of allele-specific perfect match probes, resulting in normalized intensity measurements, wherein the normalized intensity measurements are determined for the plurality of allele-specific perfect match probes without utilizing any data derived from mismatch probes;
determining and applying a cross-talk correction without use of mismatch probe data, by performing a rotation transformation to obtain cross-talk corrected normalized intensity measurements of the plurality of allele-specific perfect match probes, wherein the cross-talk corrected normalized intensity measurements include a cross-talk component from another allele that is determined to correct for probes for one allele of an SNP hybridizing with another allele of the SNP;
calculating, by the processor, an initial copy number estimate for each of the plurality of genomic regions, wherein the initial copy number estimate is based upon the normalized intensity measurements and a reference sample;
performing, by the processor, data smoothing on the initial copy number estimates, wherein the data smoothing reduces noise within the initial copy number estimates to generate smoothed copy number estimates; and
estimating, by the processor, the copy number of each of the plurality of genomic regions using a Hidden Markov Model to assign the smoothed copy number estimates to different copy number states.
18. The computer-implemented method of claim 17 , wherein the one or more user-defined metrics includes at least one of a length and a GC content of the nucleic acid molecule on which the at least one SNP resides.