Deep Sequencing Data Analysis: Challenges and Solutions

Abstract

Ultra high throughput sequencing, also known as deep sequencing or Next Generation Sequencing (NGS), is revolutionizing the study of human genetics and has immense clinical implications. It has reduced the cost and increased the throughput of genomic sequencing by more than three orders of magnitude in just a few years, a trend which is guaranteed to rapidly accelerate in the near future (Metzker, 2010). Using deep sequencing, for example, it is now possible to discover novel disease causing mutations (Ley et al., 2008) and detect traces of pathogenic microorganisms (Isakov et al., 2011). For the first time, research fields such as personalized medicine for patient treatment are becoming tangible at genomic levels given advances in deep sequencing data integration. The amount of data produced by a single ultra high throughput sequencing run is often tremendous and can reach hundreds of millions of reads in various lengths per experiment (Mardis, 2008). The storage, processing, querying, parsing, analyzing and interpreting of such an incredible amount of data is a significant task that holds many obstacles and challenges (Koboldt et al., 2010). In this chapter we will address some of the possibilities, potentials and questions raised during ultra high throughput sequencing data analysis. We will mainly focus on common pre-analysis concepts and crucial advanced considerations for alignment, assembly and variation detection. Currently, the deep sequencing user is faced with an abundance of deep sequencing data analysis tools, both publicly and commercially available. For each of the aforementioned analysis types, we will point out the various aspects to be considered when choosing a tool, and emphasize the relevant challenges and possible limitations in order to assist the user in picking the most suitable one. Since deep sequencing data analysis is a rapidly evolving field, our focus will be on fundamental concepts of the analysis process and the its challenges, allowing this read to be relevant amid additional published software. Our first part will encompass a brief overview of current leading deep sequencing technologies with special attention to their features, strengths and possible drawbacks in regards to the different preliminary questions that one might ask when using ultra high throughput sequencing. The second part of the chapter introduces pre-analysis processes. These are common quality control and assurance methods that alleviate deep sequencing derived biases and improve the overall results of any down-stream analysis. In the third part of the chapter, we will go over the different aspects of the post-sequencing analysis,