Robust Fussed Lasso Model for Recurrent Copy Number Variation Detection Conference Paper uri icon

abstract

  • 2018 IEEE. Copy number variations (CNVs) play a role in the development of several diseases, including cancer. The detection or recurrent CNVs enables us to study the regions in which they occur and understand their contribution to the formation of disease. Microarray technologies, and Array-based comparative genomic hybridization (a C GH) in particular, have been widely used in the detection of CNVs. However, due to inter-sample variability and high noise levels, simple pattern detection methods experience significant challenges in recovering the recurrent CNV regions. In this work, we propose a new method for detecting recurrent CNV regions. To achieve this goal, we propose a matrix decomposition method in which the observed aCGH probe values are estimated using two elements: I) we use a full-rank matrix of weighted piece-wise generator signals to recover the recurrent CNVs. Ii) We use a Gaussian matrix combined with a sparse matrix to capture the different types of noise and outlier values. We then evaluate the ability of our method to detect recurrent CNVs from several noisy simulated and real datasets. The results showed that our method is able to detect recurrent CNVs more accurately than current methods. Our method returned clean signals, exhibiting robustness to noise and outlier probe values.

name of conference

  • 2018 24th International Conference on Pattern Recognition (ICPR)

published proceedings

  • 2018 24TH INTERNATIONAL CONFERENCE ON PATTERN RECOGNITION (ICPR)

author list (cited authors)

  • Alshawaqfeh, M., Al Kawam, A., & Serpedin, E.

citation count

  • 0

complete list of authors

  • Alshawaqfeh, Mustafa||Al Kawam, Ahmad||Serpedin, Erchin

publication date

  • August 2018