When compared with Sanger sequencing, the major advantage offered by NGS is the ability to produce an enormous volume of data [7] at a reduced cost per base. This particular feature has expanded the applications of DNA sequencing to allow the resequencing of human genomes in a cost-effective manner. NGS, as opposed to Sanger sequencing, is based on the clonal sequencing of all the nucleic acid molecules present in the sample.

NGS technology can be achieved in a variety of commercial platforms, but they all include the following steps or methods: library and template preparation, sequencing, and data analysis. The library and template preparation step generally involves randomly breaking genomic DNA into smaller sizes or using PCR to amplify short genomic regions of interest within a sample. These short DNA fragments are then typically attached or immobilized to a solid surface or bead. The immobilization of spatially separated short DNA fragments allows millions of sequencing reactions to be performed simultaneously. The clonal amplification of templates can be achieved by solid-phase amplification [8] or by emulsion PCR (emPCR) [9]. Currently, there are several instruments being used in clinical settings that employ either solid-phase amplification or emPCR, while novel technologies are still in a nascent state. During emPCR, for example, after the successful amplification and enrichment of emPCR beads, millions are then immobilized on a chip, manufactured on wafers [10], containing millions of microscopic wells, each one designed to hold only one bead. The sequencing reaction then occurs and the addition of each new base can be detected by the release of a H⁺. On the other hand, those instruments using solid-phase amplification base the detection of the sequencing reaction by the emission of fluorescence from fluorescently labeled dNTPs. Due to the clonal nature of the NGS technology, the result is a population of template molecules, each of which has undergone the sequencing reaction multiple times. The number of times a molecule is sequenced is represented by the number of reads obtained at the completion of the sequencing run. Such reads are then aligned to a known reference sequence, such as the assembly of the human genome from February 2009 (hg19, GRCh37 Genome Reference Consortium Human Reference 37), which is the most utilized nowadays. This leads to the subsequent identification of variants, or bases that differ from the reference genome used in the alignment process. The number of reads covering a particular base in the genomic sequence is referred as the depth of coverage for that base. Typically, one can obtain a short sequence reads of up to 300–400 bases, with a depth of coverage of 30×, or up to 1,000×, depending on the application. Visualization of the raw alignments generated by this technology can be assessed by using publicly available tools, such as the Broad’s Integrative Genomics Viewer (IGV) [11] (Fig. 5.2). Much higher depth of coverage may be needed for more sensitive applications, such as detecting cell-free circulating tumor DNA in liquid biopsies, which is known as ultra-deep sequencing.

When applying NGS-based assays to variant identification in solid tumor specimens, one must take into account the fact that DNA isolated from fixed samples can be degraded and cross-linked due to formalin fixation, mixed with DNA from nonmalignant cells, or from non-mutant tumor cells, due to tumor heterogeneity, and of very small quantity, due to the increase use of less-invasive procedures, such as fine-needle aspiration (FNA) procedures. Therefore, somatic base calling can be a complex process because the mutant allelic fraction can range between 0 and 1, rather than the fixed 0.5 or 1 expected germline allelic fraction for heterozygous or homozygous mutant, respectively. Thus, in order to be able to detect 5 % mutant alleles, one would need a minimum depth of coverage of 400× [12], which is easily attainable with current NGS platforms.

One of the features of NGS that aide in reducing the cost per base is the fact that multiple samples can be sequenced simultaneously in a single run or reaction by virtue of “barcoding”. Unique sequence tags, or “barcodes” are incorporated into individual patient specimen-derived DNA during the library preparation process. After the sequencing reaction is completed, each individual sample can be sorted out during the bioinformatics pipeline. Such pipeline will align or map the reads of each individual patient sample to the human reference genome sequence. This leads to the subsequent identification of variants, or bases that differ from the reference genome used in the alignment process. A variety of different algorithms have been developed to perform the alignment and the variant calling process, with most of them being platform-specific. Finally, identified variants are further annotated by querying a variety of publicly available databases [13] to provide meaningful reports with clinically relevant information [14] (Fig. 5.3).

It is noteworthy to differentiate the detection of any mutation in a genomic region or gene with assays design to identify particular mutations of mutations that have previously been defined. Even though one could use the DNA sequencing methods described above, the detection of defined mutations can be achieved by alternative methodologies that are often faster and operate at lowers costs than DNA sequencing. The major limitation of such approaches is that they will only detect the intended mutation. Our knowledge of somatic and actionable variants in solid tumors in particular and in cancer in general is actively growing, and the need to identify multiple mutations in a single assay in small samples is becoming common practice. However, some laboratories may decide to focus their testing to a manageable number of fast mutation detection assays, at low cost. Such methods of detection are discussed in the following sections.

Typically, such methods are based on PCR applications, where primers, and sometimes probes, are designed to specifically detect the sequence of interest within the human genome. The challenge resides in the assay design to ensure sequence specificity for the detection of the intended variant and to avoid nearby polymorphisms that can prevent primer or probe hybridization resulting in no amplification or a false negative result. Such phenomenon is described as “allele dropout”, which happens when some alleles do not amplify but internal controls do. During the assay design process, one can research the sequence of interest for the presence of know single nucleotide polymorphisms (SNPs) in publicly available databases, such as the National Center for Biotechnology Information (NCBI) Short Genetic Variations database (dbSNP) [15].

Molecular methods for the identification of somatic mutations are often needed in situations where there the target is not a unique SNV, but a variety of different SNVs within a codon or exon that have similar diagnostic or predictive value. In such scenarios, the molecular assay should be designed to detect potentially unknown variants within a genomic region. This can be achieved by the detection of heteroduplex formations during the hybridization of partly complementary DNA strands, or by DNA sequencing methods.