Wabico


Wabico	Wabico : Multi-resolution GC bias correction and its application to copy number alteration identification

News	2018/02/25: Source codes are updated.
Introduction	Whole genome sequencing (WGS) data are affected by various sequencing biases such as GC bias and mappability bias. These biases degrade the detection performance of genetic variations such as copy number alterations. Previous methods use relation between GC proportion and depth of coverage (DOC) of markers using various regression models. But the severities of the GC bias vary sample to sample. We developed GC bias correction method with multi-resolution analysis. We used translation-invariant wavelet transform to decompose raw biased signal into high and low frequency coefficients. Then we modeled relation between GC proportion and DOC of the genomic regions and constructed the artifical control DOC. When we applied our method to the simulated sequencing data with various degrees of GC bias, it shows more robust GC bias correction performance than the competing method. We also applied our method to the public cancer sequencing datasets to identify copy number alterations and found most cancer related focal alterations without using normal control sequencing data. Our method can be applied to various WGS data with different degrees of GC bias.
Paper	Jang, H. and Lee, H. (2018) Multi-resolution GC bias correction and its application to copy number alteration identification
Source codes	Source codes are available here. Manual pages for Wabico is here.
Contacts	hyunjulee at gist.ac.kr

Whole genome sequencing (WGS) data are affected by various sequencing biases such as GC bias and mappability bias. These biases degrade the detection performance of genetic variations such as copy number alterations. Previous methods use relation between GC proportion and depth of coverage (DOC) of markers using various regression models. But the severities of the GC bias vary sample to sample. We developed GC bias correction method with multi-resolution analysis. We used translation-invariant wavelet transform to decompose raw biased signal into high and low frequency coefficients. Then we modeled relation between GC proportion and DOC of the genomic regions and constructed the artifical control DOC. When we applied our method to the simulated sequencing data with various degrees of GC bias, it shows more robust GC bias correction performance than the competing method. We also applied our method to the public cancer sequencing datasets to identify copy number alterations and found most cancer related focal alterations without using normal control sequencing data. Our method can be applied to various WGS data with different degrees of GC bias.