Somatic mutations are mutations acquired by non-germline cells and cannot be inherited by the offspring of the parent organism of the mutated cell, with the exception of, for example, canine transmissible venereal tumor [6]. Somatic mutations are important in the diversity of the antibodies, T cell receptors, and B cell receptors. They are frequently caused by environmental factors and accumulate in the DNA of any organism despite proficient DNA repair mechanisms. Somatic mutations are present in healthy tissues at a frequency of about 2-6 mutations per million bases [7], and about three somatic mutations per healthy human individual [8]. As a consequence, somatic cells present different genotypes within the same individual, a widespread phenomenon in healthy development and aging, known as somatic mosaicism [9, 10]. Somatic mutations accumulate during the aging process. Endogenous somatic mutations are a contributing factor in aging, since the rate of somatic mutations across mammlian species displays an inverse relationship with species lifespan [11]. K Nanki et al measured 0.019 single nucleotide variant per Mb per year among colonic epithelia [12]. Zhang L et al, using Single-Cell Multiple Displacement Amplification whole genome sequencing, found that the number of somatic mutations in human B lymphocytes increased from <500 per cell in newborns to >3,000 per cell in centenarians [13]. K Yoshida et al estimated 22 single-base substitutions per cell per year among the bronchial epithelial cells [14].

The implications of mosaicism in disease are not completely understood [15, 16], but recent studies suggest that it influences cancer [17-21], neurological and neuropsychiatric disorders [16, 22-25], ulcerative colitis [12], and various monogenic disorders [26]. In cancer cells, such mutations were found at high frequency. In a review by Martincorena and Campbell (2015) [18] it is stated that between 1000 and 20,000 point mutations and hundreds of insertions, deletions and rearrangements can be detected in cancers. These numbers were determined based on studies on millions of mutations in various cancer types [27-30]. COSMIC (the Catalogue Of Somatic Mutations In Cancer) [31] version 91, released in April 2020, includes almost 34,657,730 coding mutations across 1,443,198 samples, curated from 27,496 papers . The catalog is a great resource for cancer research, for example in [32, 33]. TCGA is another resource with somatic mutations in various cancers [32]. Somatic mutations on ENL likely contribute to Wilms tumour machanistically [34]. Researchers have started treating patients based on somatically mutated genes, for instance, a patient has been successfully treated with tumor-infiltrating lymphocytes reactive against mutant SLC3A2, KIAA0368, CADPS2 and CTSB [35].

Somatic mutations are not reduced to point mutations but can be any genomic variation: repeats, deletions, insertions, multiplication, loss of copy number, and others. Chromosomal somatic mutations occur when somatic cells divide. During this time, structural aberrations result from chromosome breakages and incorrect repairing or the unequal exchange of material during chromosome separation. Structural aberrations include deletions, when a part of the chromosome is missing; duplications when a portion of the chromosome appears twice; translocation when genetic material has been interchanged between non-homologous chromosomes or inversions, when part of the chromosome is in inverse orientations. Since such changes occur only in some cells, chromosomal mosaicism is observed.

One common concern when trying to identify and characterize somatic mutations is their low abundance. Due to the low frequency, somatic mutations are difficult to detect in bulk tissue samples, and amplification of template DNA is necessary before analysis. However, DNA amplification is an error-prone process, and real mutations may be hidden among these errors. Even the modern sequencing technologies fail to detect many somatic variants, especially the ones present in very few or single copies. The sequencing errors are high enough to mask real variations [36]. Sensitive methods are necessary for the detection and relative quantification of somatic mutations in biological material.

Due to the variety of mutation types and the difficulties mentioned above in detecting somatic mutations, a wide range of techniques have been and continue to be developed. PCR, electrophoretic, chromatography and sequencing methods are most commonly used for detecting either known (for diagnosis) or unknown point mutations. To these, probing methods are added for the detection of chromosomal aberrations. Detection methods can be specific to targeted chromosomal regions or used for whole exome, for example, in the study of canine transmissible venereal tumor [6] or colonic epithelium associated with ulcerative colitis []31853059, whole genome or transcriptome analysis.

Here we present classical and modern methods for identification of somatic variations. Each method is briefly described, and studies that successfully used it are exemplified. The list includes commonly used techniques, without being considered exhaustive. As new technologies are being developed, they are added to this collection. Table 1 provides a summary discussing usefulness, sensitivity and some advantages and disadvantages.

Variations of PCR have been used to identify known small mutations.

Single point or small insertion/deletion mutations are commonly detected by the Amplification Refractory Mutation System (ARMS) technique and its variations. This method, also know as allele-specific PCR (ASP) or PCR amplification of specific alleles (PASA), is a conventional PCR-based method used to detect single base mutations in a complex pool of DNA molecules. In this method, mutations at known locations in the target sequences can be identified by performing two complementary PCR reactions with one common 5’- primer specific to the target DNA region, and two 3’- primers that are allele-specific. The only difference between these two primers is at a given base, such that one complements the wild-type sequence and the other the mutant. If the sample is homozygous amplification will only occur in one of the tubes; if the sample is heterozygous amplification will be seen in both tubes (Figure 1A). One essential factor in this method is the thermostable polymerase that lacks the 3’-5’ exonuclease activity, like Taq polymerase, used to avoid the elimination of the mutation in the mutant-specific primer. Agarose gel electrophoresis followed by ethidium bromide staining is commonly used to visualize the PCR products. This is a rapid and efficient method when the PCR conditions are properly chosen [38, 86, 87] and can also be used for amplification of target sequences containing small insertion or deletions.

Fluorescent labeling and DNA sequence analysis improve the technique, allowing for co-amplification of normal and mutant DNA fragments with different size primers [88, 89]. Relative quantification of allele proportions is possible by combining ARMS with analysis of the DNA product melting-point temperatures obtained from post-PCR fluorescent dissociation curves (DCA) [42, 90, 91]. A further improvement of the technique is seen in tetra-primer ARMS (T-ARMS) in which two common primers are used to amplify a DNA sequence as an internal control, while opposite orientation allele-specific primers help amplify size –specific fragments (Figure 1B) [40, 92-94]. Other variants of ARMS include competitive oligonucleotide priming (COP) [95], mutant enrichment PCR (enriched-EPCR or mutant-enriched - ME-PCR) [96, 97], mismatch amplification mutation assay (MAMA) [98], mutant allele-specific amplification (MASA) [99], and allele-specific competitive blocker-polymerase chain reaction (ACB-PCR) [100].

[enlarge]

Figure 1. Conventional PCR based allele-specific amplification and mutant DNA detection. A) Steps of ARMS amplification of DNA at a known location; B) Tetra-primer ARMS variant technique.

A combination of allele-specific PCR and real-time PCR can also be used to detect minority alleles [42, 43, 101-104]. In real-time PCR the amount of PCR product is measured after each round of amplification using a fluorescent readout [105]. The amount of target sequence is determined based on a standard curve obtained from samples of known copy number, assuming that the amplification efficiencies of the sample and the standards are equivalent.

As in real-time PCR, the amplified material in reverse-transcription PCR is mRNA. In this technique, the mRNA in fist transcribed into a complementary DNA strand, which is subsequently amplified in a classical PCR amplification cascade. The final PCR product is sequenced by Sanger sequencing. This method proved useful in detecting unexpected RNA splice variants that contained various mutations (intra-exonic junctions, insertions, deletions, single nucleotide variations) in a recent study on small populations of neurons from Alzheimer’s disease patients [45]. In combination with other observations (see below), this finding led to the important conclusion that neuronal gene recombination may lead to the presence and the expression of short variants of normal genes, which might be a mechanism with significant consequences for the normal functioning and occurrence of diseases in the human brain.

This technique is ideal for the study of somatic mutation at the level of RNA as well as for detection of chromosomal translocations [44, 45, 106].

In digital PCR the sample is diluted such that every dilution contains a minimum amount of target sequence, ideally one or none. Each dilution is used as a template for a PCR reaction, and the amplified product is detected by fluorescence. The distribution of target sequences in the diluted samples can be approximately determined by a Poisson’s distribution, while their concentration can be calculated based on the ratio of samples with positive amplification to total dilution samples. In dPCR, the fluorescent signal is measured at the end of the amplification and does not rely on comparisons with a standard curve. This way, the measurement is not dependent on the efficiency of the amplification reaction, as is the case in qPCR. Multiple somatic mutations have been identified as causes for several malformative disorders [107, 108], by using this method.

A comparison of the conventional, real-time and digital PCR techniques is shown in Figure 2.

[enlarge]

Figure 2. Comparison of PCR-based techniques (from Quan et al, 2018 [1] ).

Other variations of the PCR technique use modified nucleic acids, like peptide nucleic acids (PNAs) [50, 109, 110] or locked nucleic acids (LNAs) [53, 54, 111], as a substitute for primers. In these techniques, the wild-type DNA amplification is suppressed, and the mutant template is being amplified. Similarly, a DNA tag containing a highly thermostable restriction nuclease recognition site has been used in FLAG (fluorescent amplicon generation) assays that combine PNA probes and restriction endonuclease-mediated selective (REMS) PCR (Figure 3) [2, 48]

[enlarge]

Figure 3. FLAG assay and PNA-PCR based selection of mutant templates. (Modified from Amicarelli et al, 2007 [2]. A) FLAG assay principle. A restriction nuclease sensitive site (PspGI) is inserted by DNA amplification only when the DNA sequence is wild-type. One of the primers is labeled with both a quencher (Q) and a fluorophore (F). When wild-type DNA is amplified, digestion leads to a fluorescent signal increase due to loss of quenching. B). PNA-mediated FLAG assay. When a PNA specific to a particular mutation is added to the reaction, the formed PNA/DNA hybrid inhibits primer annealing and DNA amplification such that no fluorescent signal is obtained. However, primer hybridization is not inhibited by incomplete complementarity of the PNA (single mismatch) leading to generation of a fluorescent signal. By conducting four distinct FLAG reactions with PNA probes specific for different mutations (GTT, GAT, GCT, TGT), the identity of the mutation can be inferred from the intensity of the fluorescent signals obtained.

The COLD-PCR techniques is a single-step technique that permits the preferential amplification of DNA fragments with known or unknown mutations, based on the lower annealing temperature of mismatch-containing duplexes. Mutated DNA sequences are enriched during PCR by lowering the denaturing temperature such that imperfect pairs are denatured but not the fully complementary wild-type duplexes. A new round of primer annealing now has mostly mutant DNA available for pairing and replication [55, 112, 113]. Two forms of the technique have been developed. One permits the enrichment of all mutants (Full COLD-PCR – Figure 4 A) [114, 115], while the second only allows the amplification of melting temperature-reducing mutations (Fast COLD-PCR - Figure 4B) [116, 117].

[enlarge]

Figure 4. Principle of COLD-PCR. A) Full COLD-PCR; B) Fast COLD-PCR.

This technique allows for the differentiation of wild-type and mutant DNA strands based on their mobility shifts produced by their different conformations when single-stranded. The separation of differently structured denatured DNA strands can be done either in gel [118] or through capillaries [119, 120], when an electric current is applied (GE- gel electrophoresis or CE – capillary electrophoresis SSCP). Figure 5 illustrates the principle of this technique.

[enlarge]

Figure 5. Single-strand conformation polymorphism detection by gel electrophoresis (A), and capillary electrophoresis (B). In capillary electrophoreses fluorescently labeled DNA fragments are denatured, separated according to their migration, and detected based on their fluorescent signal.

Denaturing gradient gel electrophoresis (DGGE) allows the screening of PCR-amplified DNA for single base mutations. In this technique, a mixture of DNA fragments of different sequences is separated by electrophoresis in an acrylamide gel containing a linearly increasing gradient of denaturant. More stable DNA fragments like the G-C rich fragments migrate faster, while denatured molecules move slower in the gel. In this manner, DNA fragments can be separated in the acrylamide gel. DNA fragments can be extracted from the gel, amplified and sequenced [61]. Tumor-specific mutations have been detected successfully by this method [121].

CDCE is another technique that uses the power of denaturants, thermal or chemical, to separate different sequence DNA fragments through capillaries under electric currents. The method can be used to rapidly detect point mutations in candidate disease genes [122-125].

Separation of nucleic acids based on their size can be easily achieved by chromatography. In addition, when denaturants are added, the molecules can be separated based on their structure.

Similarly to the other techniques that use denaturants to detect conformational changes in DNA molecules with a slightly different sequence, liquid chromatography can be used for separating such molecules based in differences in their column retention time [67, 126-130].

Enzymatic mutation detection takes advantage of natural processes in which DNA-cutting enzymes are involved.

Several methods for detecting single nucleotide changes in DNA exploit the ability of the mismatch repair proteins to bind mismatched nucleotide pairs. One of the commonly used enzymes, the E. coli MutS can bind heteroduplexes with up to four mismatched pairs. The enzyme has been immobilized on a variety of solid supports (nylon, nitrocellulose, PVDF membrane) [131, 132] or to protein chip matrices [133, 134] and used to scan amplified DNA for fragments containing mismatches in vitro.

Moreover, in vivo detection of mismatch-containing duplexes has been developed. Faham and Cox (1995) [135] used E. coli to screen DNA fragments for variations. They cloned DNA fragments into two MRD plasmids of which one can, and one cannot express lacZ due to a 5bp insertion. They amplified the plasmids into both methylation deficient and wild-type bacterial strains. Extracted methylated and unmethylated plasmids were mixed, denatured, renatured, and subjected to digestion with specific restriction endonucleases. Since only fully methylated or unmethylated DNA is degraded, hemimethylated DNA is selected. This heteroduplexed plasmid is transformed in E. coli where it triggers the mismatch repair response, resulting in a repaired lacZ gene. A white colony grows and can be selected. In the absence of a mismatch, DNA replication produces both lacZ variants, and blue colonies grow. The method does not allow for the identification of the particular variations but can find all DNA fragments that contain mutations, and it can be adapted for scanning large genomic regions [70, 136-138].

Mutant enrichment has been achieved in restriction endonuclease-mediated selective PCR (REMS-PCR) with the help of heat-resistant restriction enzymes that selectively destroy wild-type [2, 71, 104, 139] or mutant [140] DNA during PCR. A combination of restriction enzyme digestion and real-time PCR (real-time digestion PCR) proved to be efficient in detecting somatic mutations [141] Alterations in restriction enzymes’ recognition or cut sites as a result of point mutations can also be identified as a result of altered restriction digestion patterns in restriction fragment length polymorphism (RFPL) [142-145].

High-resolution melting (HRM) curve analysis represents a fast, post-PCR high-throughput method for scanning somatic sequence alterations in target genes. In this method, the fluorescent signal from intercalating dyes in dsDNA is read as a measure of the degree of hybridization. dsDNA is denatured gradually by increasing temperature, and the decrease of the fluorescent signal is measured generating “melting” curves. Shifts of the melting curves permit the identification of mutant DNA fragments (Figure 6). The method has been successfully used to detect somatic mutations in multiple types of cancer [72, 146-149].

[enlarge]

Figure 6. High-resolution melting (HRM) curve analysis of wild type and mutant DNA. A shift of the melting curve indicates a mutant DNA fragment.

A novel technology developed by Sequenom (www.sequenom.com) - the Mass ARRAY system - combines PCR amplification, single-base primer extension and mass spectroscopy of DNA for mutation detection. Wild type and mutant DNA is amplified by multiplex PCR, and the primer extension reaction is performed with mass-modified terminator nucleotides. The product is analyzed, and mutant and wild type fragments are identified by MALDI-TOF (Matrix-Assisted Laser Desorption/Ionization – Time of Flight) [74, 75, 150].

DNA microarrays are solid support grids of fixed short ssDNA fragments (oligonucleotides) that act as probes for tested genomic DNA targets. Highly complementary DNA hybridizes with the probe even under stringent conditions, while mismatched pairs are less stable. Probe-target hybridization is usually detected and quantified by detection of fluorescent, chemiluminescent or colorimetric signal. The intensity of the signal from each spot on the grid is determined under different hybridization conditions, and the relative abundance of each nucleic acid sequences in the target is identified, thus allowing the identification of subpopulations of sequences, like somatic mutations-baring fragments with low error frequency [151, 152]. This method allows the detection of low abundance point mutations in complex samples [76].

In Multiplex Ligation-dependent Probe Amplification (MLPA) probe pairs hybridize specifically head-to-tail to target sequences in liquid media. Hybridized probes are subsequently ligated and amplified by PCR using fluorescent primers complementary to tag-sequences inserted in the probes (Figure 7A). PCR products are separated by capillary electrophoresis on an automated fragment analyzer, where peak heights indicate the amount of amplified product of each separate probe pair (Figure 7B). The absence or the poor amplification of specific probes means deletions, while unusually high amplification suggest gene duplications [153-157]. MLPA is a high throughput and cost-effective method that has been used in multiple studies of which only a few are referenced here [158-162].

[enlarge]

Figure 7. Multiplex Ligation dependent Probe Amplification A) Schematic diagram; B) Typical capillary electrophoresis automated fragment analyzer result (modified from [3] ).

In situ hybridization (ISH) can be used to determine the exact order of specific DNA and RNA fragments within a larger molecule. The technique is based on the ability of the double-stranded DNAs to denature to single-stranded upon heating and to pair with its complementary DNA or RNA sequence under non-denaturing conditions. When a labeled fragment of DNA (a DNA probe) is denatured and added to denatured DNA or RNA samples, some of the labeled DNA will hybridize to its complementary sequence. These duplexes are detected based on the detection of the probe, commonly by detection of its fluorescent signal. Two probes specific to sequences distant in the genome, which anneal at the same site in a chromosome, indicate a possible DNA junction. Based on this technique, exon-exon junctions were detected in short mRNA and gene variants of the APP gene associated with Alzheimer’s disease [45]. Such recombination events are somatic, and occur only in diseased neurons, but not in normal brains [45]. DNA sequencing of the probe-detected regions can confirm the junction.

All somatic mutations can be identified by determining the complete nucleotide sequence of the genome. Several methods have been developed, starting from the first generation (classical) Sanger sequencing, to the more sophisticated next-generation approaches.

Direct sequencing is based on Sanger’s chain termination reaction technique [163]. It is a simple, accurate, and practical method for identifying mutations in genes associated with disease when the number of samples available is small. For example, multiple mutations were identified in the TCOF1 gene associated with Collins Treacher syndrome [164], mutations in the FGFR2 gene were detected in patients with Apert syndrome [165], while mutations in the PI3K [166] and KRAS [167] genes were identified in gallbladder and colon cancers using direct sequencing. More recently, direct sequencing was used in combination with locked-DNA PCR (LNA-PCR) to detect low-frequency mutations. Albitar et al [168] identified 1 mutant in a background of 1,000 wild type alleles by this method.

In spite of its simplicity, the method is not appropriate for the detection of unknown mutations when a large number of genes are candidate genes or when there is no candidate gene at all. It is also not considered the method of choice for detecting small deletions, but it is practical for the detection of mutations in genes involved in genetic disorders.

In contrast to the direct sequencing method devised by Sanger, next-generation sequencing involves the repeated sequencing of DNA segments randomly generated by genome fragmentation. The resulting sequence reads, of which many partly overlap, are aligned to a reference sequence (if available) and the full genome sequence is determined based on consensus [169, 170]. A few platforms for NGS have been developed and are used for both DNA and RNA sequencing. These include Illumina (Solexa) sequencing, Roche 454 sequencing, and Life Technologies Ion torrent: Proton / PGM sequencing or SOLiD sequencing, and were reviewed by Gundry and Vijk, 2012 [171, 172]. Alternatively, the full exosome [173, 174] or just specific targets are being sequence analyzed, especially for identification of mutations in monogenetic disorders [175]. The Agilent Sure Select Target enrichment system has been recently used for the detection of APP short gene variant in brains with Alzheimer’s disease [45].

All platforms allow the fast sequencing of entire genomes but are error-prone [36]. Because sequencing errors might not be distinguished from true mutations in the case of low-abundance somatic mutations, approaches to increase the sensitivity of the sequencing instruments were developed. Among them are Duplex Sequencing [176], Safe-Sequencing System ("Safe-SeqS") [36] and circle sequencing [4].

In duplex sequencing, the two strands of a DNA duplex are sequenced independently. As the two strands are complementary, true mutations are found in the same position in both strands [176]. Safe-SeqS is a barcoding method that involves the individual labeling of each DNA fragment, followed by amplification and the sequencing of the amplified product. Since multiple daughter molecules with identical sequence tags are generated, a preexistent mutation will be present in every daughter molecule containing the same tag in contrast to mutations introduced by sequencing or amplification which could be present in a much smaller subset of molecules [36]. In circular sequencing, DNA is denatured to single-stranded and circularized before performing rolling circle replication. Tandem-linked copies of the template circle are sequenced using any high-throughput sequencing technology, and the read sequence is computationally divided into individual copies of the original circle. If the first circle contains a mutation, the mutation will appear in multiple copies [4]. Figure 8 shows a comparative illustration of the above-described methods.

[enlarge]

Figure 8. Overview of barcoding, circle and duplex sequencing (modified from [4].)

The several approaches available for analyzing somatic mutations apply mostly to point mutations and small insertions or deletions. Large deletions, insertions, inversions, or translocations cannot be detected by bulk DNA sequencing. An alternative sequencing method, ultra-low coverage sequencing, can be used to identify large somatic mutations by using a single sequencing read that spread across multiple reads as unique events [177].

In this technique for DNA sequencing, a single DNA polymerase or reverse-transcriptase enzyme and a single molecule of template DNA or RNA are localized at the bottom of an optical unit called zero-mode waveguide (ZMW). The ZMW contains a detector able to detect base-specific fluorescent signals from each nucleotide incorporated. The four bases are pre-labeled with specific fluorescent dyes, and upon their incorporation into the nascent DNA the tags are released. They diffuse out of the ZMW’s area where fluorescence is no longer observable. Thus, DNA synthesis can be observed in real-time. The technique developed by PacBio ( Smrt-seq was recently used by Lee et al to detect somatic variants of Alzheimer’s disease-associated gene APP [45]. In this study, both amplified DNA and cDNA fragments, as well as APP specific RNA were used as templates. Vasan N et al used PacBio SMRT-seq to resolve whether two PIK3CA mutations were on the same allele (cis configuration) or different alleles (trans configuration) [183].

Sequencing the genome of individual cells can reveal somatic mutations and facilitates the analysis of clonal evolution. To perform any single-cell sequencing assay, individual cells first have to be isolated from the system of interest. The most commonly used method is FACS. However, serial dilutions, microfluidic devices [184], optical tweezers [185, 186], and manual micromanipulation [187, 188] can also be used. All these methods have been reviewed elsewhere [189-193]. Methods for isolating single cells from rare cellular populations have also been developed and were reviewed recently by Navin NE (Figure 9) [5].

[enlarge]

Figure 9. Summary of single-cell sorting methods reviewed by Navin, NE [5]. A) Methods for isolating cells from abundant-cell populations. B) Methods from isolating cells from rare-cell populations.

Rare cells have been enriched from circulating tumors by CellSearch (Johnson and Johnson) [194], isolated by DEPArray (Silicon Biosciences) [195], and their whole genomes were amplified and analyzed by NGS (Next-Generation Sequencing) [196, 197] or by Sanger sequencing [198, 199] to identify somatic mutations in the genome. Other studies have used whole genome amplification after single cell isolation by laser microdissection [178, 200-202] and different sequencing strategies [200, 201, 203]. Some skipped the whole genome amplification step altogether [204]. A different technology, called MagSweeper (Illumina) that uses a rotating magnet with bound EpCAM antibodies to isolate circulating tumor cells [205] was used in combination with DNA sequencing [206] to detect somatic mutations in tumor cells in breast cancers.

An alternative to genome sequencing for somatic mutation detection from single cells is the sequencing of the cell’s transcriptome through RNA-seq, as it can provide information at single nucleotide resolution. Mutations can be identified directly by RNA-seq [207, 208] or validated after their detection by DNA-seq or other methods [209]. RNA-seq is often the sequencing method of choice due to budget limitations, sample quantities or study goals. However, analyzing RNA-seq data can be challenging because it can have a high false positive rate for single nucleotide variations [210, 211]. This is a consequence of the biology of RNA metabolism (e.g., RNA editing, RNA splicing), as well as of the molecular techniques (e.g., errors introduced during reverse transcription and PCR) and sequence analysis methods (e.g., strand bias [211, 212], alignment complexity [213] ).

Most RNA-seq data specific variant detection tools were developed for single nucleotide variation detection rather than for somatic mutations [213, 214], but more recent devices are designed and used to identify somatic mutations [208, 215-217].

Yizhak K et al adapted the MuTect platform [218] to identify somatic mutations from large RNA-seq databases Cancer Genome Atlas and Genotype-Tissue Expression (GTEx) project, validated their approach through Fluidigm microfluidic PCR and the Illumina MiSeq sequencing system, and identified skin, lung, and esophagus as most frequent tissues for somatic mutations [219].

Somatic mutations can be identified from parallel sequencing data by directly comparing the DNA sequence from tumor samples with their normal counterparts. The paired tumor/normal approach permits the identification and elimination of genomic variants due to their presence in all the cells of an individual. Somatic mutations are identified based on their existence only in a subset of cells. For example, T Powles et al identified somatic variants based on the tumour tissue and matched normal sequencing data and used those variants as the markers to detect the presence of circulating tumour DNA [83]. Commonly mutated sites have been identified in a variety of cancers by using this method [220]. One inconvenience of this approach is the need to sequence twice as many samples as the number of test samples, leading to higher than needed costs and analysis duration. Also, for each test sample, a matching standard tissue sample is required, and sometimes this might not be available.

Recently, an alternative approach, in which only test samples are used, has been developed [221], but data suggest that matched tumor-normal sequencing is needed for precise identification of somatic mutations [222]. A slightly improved normal-tissue free analysis technique was developed by Jamie et al [223]. This method uses a specialized Genome Analysis Tool Kit (GATK) pipeline [224] to improve the detection of somatic mutations.