Jitter representation in SCNA breakpoints using asymmetric exponential power distribution

Jitter is inherent to the breakpoints of measured genome somatic copy number alterations (SCNAs). Therefore, an analysis of jitter is required to reduce errors in the SCNA estimation. The high resolution technologies of hybridization are used to detect SCNAs. However, the SCNA measurements are accompanied with intensive noise that may cause errors and ambiguities in the breakpoint detection with low signal-to-noise rations (SNRs). In this paper, we show that the asymmetric exponential power distribution (AEPD) provides much better approximation to the jitter distribution than the earlier proposed discrete skew Laplace distribution. We use the AEPD to approximate the jitter distribution by finding the best fit for the measured SCNAs. The proposed approximation is tested experimentally by data generated with several values of the SNRs.

I. Introduction

The chromosomal changes called the copy number alterations (CNAs) are gains or losses of large segments in the deoxyribonucleic acid (DNA) of the genome, which are essential for all known forms of life. The somatic CNAs (SCNAs), which occur during the lifetime of an individual are recognized as mayor contributors to cancer development [1]–[3]. Despite a great progress in the chromosomal probing using most modern technologies to measure the genome chromosomal structure [4], the measurement noise still remains at a high level in the SCNAs data.