Confidence limits for genome DNA copy number variations in HR-CGH array measurements

• Estimation of genome copy number variations is provided in large noise.
• The estimation accuracy is limited with jitter in the breakpoints.
• The approximate jitter distribution is shown to be the discrete skew Laplace.
• The estimate lower and upper bounds are derived.
• The estimate UB and LB masks are suggested for medical applications.

Estimation of the genome copy number variations (CNVs) measured using the high-resolution array-comparative genomic hybridization (HR-CGH) microarray is commonly provided in the presence of large Gaussian noise having white properties with different segmental variances. Medical experts must thus be highly concerned about the confidence limits for CNVs in order to make correct decisions about genomic changes. We carry out a probabilistic analysis of CNVs in HR-CGH microarray measurements and show that jitter in the breakpoints can be approximated with the discrete skew Laplace distribution. Using this distribution, we find the confidence upper and lower boundaries to guarantee an existence of genomic changes in the confidence interval of 99.73%. We suggest combining these boundaries with the estimates to give medical experts more information about actual CNVs. Experimental verification of the theory is provided by simulation and using real HR-CGH microarray-based measurements.