Abstract 1660: Identification of allelic imbalance utilizing heterozygous genotype allele frequencies and intensities

Abstract

Abstract Allelic imbalance (AI) events, such as amplification, deletion or copy-neutral loss-of-heterozygosity (cn-LOH) can result in the activations of oncogenes or inactivations of tumor suppressor genes that are critical to the process of carcinogenesis and metastasis. However, tumor samples often have low tumor-cellularity and small subclones that require sensitive and robust algorithm for AI detections. To overcome this challenge, our lab has developed an original software application hapLOH (Vattathil and Scheet, 2013) that identifies AI with high sensitivity in SNP array data by utilizing heterozygous genotype allele frequencies in SNP array data using a Hidden Markov Model. Motivated by the plethora of next-generation sequencing (NGS) data, we carried this logic and functionality and developed hapLOHseq (San Lucas et al, 2016) for detection of subtle AI in NGS data using reference and alternate allele read depths and a set of haplotype estimates based on 1000 genome data. Recently, we sought to improve AI detection and provide support for different data types by developing haploh-cn. Haploh-cn extends our previous tools by supporting both SNP array and NGS data, and by incorporating intensity data and Log R ratios (LRR) into the underlying Hidden Markov Model for SNP array data, and incorporating sequencing depth for NGS data. In order to evaluate haploh-cn’s performances, we have downloaded The Cancer Genome Atlas Lung Adenocarcinoma whole exome sequencing data and corresponding Affymetrix SNP6 array data. Then, we simulated tumor cellularity by performing in-silico dilution of NGS and SNP array data. For NGS data, we down sampled tumor sequencing BAM files and mixed in matched-normal sequencing reads. For SNP6 array data, we combined the adjusted B allele frequencies of heterozygous sites in the tumor sample and matched-normal sample. Using the AI events detected in the undiluted tumor samples as the gold standard, we assessed the performance of haploh-cn in the computational diluted samples. Our results showed that haploh-cn had high sensitivity without sacrificing specificity in diluted tumor samples. In summary, haploh-cn is a robust and powerful method for profiling subtle AI in NGS and SNP array data. Citation Format: Kyle Chang, Francis A. San Lucas, Zuhal Ozcan, Smruthy Sivakumar, Yasminka A. Jakubek, Richard G. Fowler, Paul Scheet. Identification of allelic imbalance utilizing heterozygous genotype allele frequencies and intensities [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 1660.