AluScan is a pre-sequencing capture method with reduced costs and cancerous DNA sample requirement, using inter-Alu PCR in conjunction with next-generation sequencing (NGS). As an efficient solution to challenges in current cancer genome studies, AluScan generates an Alu-anchored scan on human genome to achieve inter-Alu sequences enriched in genic regions. With the built-up pipeline and developed programs, AluScan sequences from cancer genomes were employed in analysis of genomic alterations including single nucleotide variation (SNV), loss of heterozygosity (LOH) and copy number variation (CNV) to reveal the potential underlying pathways and hence improve the treatment of the tumors.

In this thesis, two programs developed for processing AluScan sequencing data were introduce...[ Read more ]

AluScan is a pre-sequencing capture method with reduced costs and cancerous DNA sample requirement, using inter-Alu PCR in conjunction with next-generation sequencing (NGS). As an efficient solution to challenges in current cancer genome studies, AluScan generates an Alu-anchored scan on human genome to achieve inter-Alu sequences enriched in genic regions. With the built-up pipeline and developed programs, AluScan sequences from cancer genomes were employed in analysis of genomic alterations including single nucleotide variation (SNV), loss of heterozygosity (LOH) and copy number variation (CNV) to reveal the potential underlying pathways and hence improve the treatment of the tumors.

In this thesis, two programs developed for processing AluScan sequencing data were introduced, one was SAMSVM and the other was AluScanCNV.

For SAMSVM, it was developed as a tool for misalignment detection and filtration on SAM-format sequences from Illumina platform with use of support vector machine (SVM). Employing LIBSVM packages, SAMSVM performed misalignment detecting with high accuracies ranged from 0.89 to 0.97 and F-scores ranging from 0.77 to 0.94 in benchmarking of simulation data. Also, it increased mapping rate and on-target rate of SNP calling on real data.

For AluScanCNV, it was developed as a tool for CNV calling on AluScan data. Employing Geary-Hinkley transformation (GHT) and circular binary segmentation (CBS), AluScan performed localized CNV calling and extended CNV calling in practical use. The CNV calling result of liver cancers resembled the results obtained from whole genome sequencing (WGS) study. Also, the validation test on existing dataset showed high correlation (R = 0.935 in CNV loss calling and R = 0.776 in CNV gain calling) with another CNV calling tool named FREEC.

In this thesis, AluScan sequencing data from ten hepatitis B virus (HBV) positive and five non-viral hepatocellular carcinomas (HCCs) were subjected to comprehensive analysis to reveal genomic difference between viral and non-viral HCCs. Generally, non-viral HCCs displayed far fewer SNVs than HBV-positive HCCs, whereas these two types of HCCs showed similar patterns of LOH preferences and contained similar levels of CNVs in comparable genic locations. Mutational signature analysis showed that the two types of HCCs displayed specific signatures in base substitutions, suggesting that virus infection could result in specific SNVs. Signature V1 enriched in C>T mutation at NpC̲pG sites, suggesting that deamination of the methyl-5’ cytosine could be associated with virus infection; the mutually reversible T>A mutations at ApT̲pC and GpT̲pT were found in non-viral HCCs but only T>A mutations at GpT̲pT were observed in HBV-positive HCCs. In addition, results of hierarchical clustering on samples and selected functional events (SFEs) showed that non-viral HCCs belonged to C-class while HBV-positive HCCs belonged to M-C mixed class in terms of their dominant mutations. Lastly, the cancer genes that were found to contain mutations were shown to be associated with pathways of oncogenesis in both types of HCCs, while HBV-positive HCCs in addition contained mutations in genes associated with pathways of virus infection. In conclusion, the results suggest that the SNV-CNV mutational profiles of HBV-positive HCCs differ from the profiles of non-viral HCCs. These differences could be important to both the understanding and the therapeutic treatment of these two types of liver cancers.