Fig. 5

Methylation analysis and whole-genome sequencing of cfDNA enables applications that extend beyond DNA sequence and copy number. A (Left) Methylation patterns of DNA in tumor cells may look different from their normal, healthy counterparts. Generally, CpG islands are associated with promoter regions of genes and these regions are prone to hypermethylation, i.e., gain of methylation, in tumor cells, leading to a block of gene transcription, as the bulky transcription machinery is prohibited from binding to the hypermethylated site. Conversely, tumor cells exhibit a general trend of global hypomethylation, i.e., loss of methylation, throughout the genome, which is frequently observed at repetitive sequences. The lollipops represent CpG sites, with white lollipops indicating no methylation at this particular cytosine and dark blue representing methylation at the cytosine. (Right) Typically, beadchip array data can be harvested to perform differential methylation analysis between various tissue types of interest, e.g. comparing normal breast tissue and malignant breast tissue or identifying differences in methylation between healthy colon or lung tissue. Differential individual CpGs or regions of differential methylation can be identified for use as a tissue-specific marker for downstream purposes. CpGs or regions of CpGs that do not confer a highly differential methylation signal from other analyzed tissues will not constitute a robust tissue-specific marker. B Apoptotic death of cells results in the digestion of open chromatin, i.e. regions of DNA not bound to and protected by nucleosomes. Naked DNA not associated with proteins, e.g., histones or TFs, will be digested and not detected in the circulation. C The majority of cfDNA is thus mononucloeosomal DNA. However, longer fragments of DNA may be protected by two nucleosomes, i.e. dinucleosome. D The coverage patterns of where the reads align in the genome reflect the biology of that particular region. The coverage patterns at regions of interest (ROI) reflect the original positioning of nucleosomes in cells. Generally, well-defined nucleosome organization and positioning in cancer cells may indicate that the ROI is “open” or accessible. This is accompanied by a drop in coverage at the ROI, where no nucleosomes were positioned, resulting in what is referred to as the nucleosome-depleted region (NDR). Densely packed nucleosomes with less defined positioning reflects that the region is not accessible or “closed”, with no drop in coverage at the NDR and no oscillation of coverage upstream or downstream to the ROI. Example ROIs are transcription start sites (TSS), transcription factor binding sites (TFBSs), or DNase hypersensitivity sites (DHS). E The types of fragment features that can be observed are diverse, such that there is no one-size-fits-all approach to applying fragmentomics to cfDNA. Exemplary features that can be harvested for analysis are illustrated on this DNA strand, including fragment length. Green stars represent that plasma DNA ends show prevalence of certain nucleotide contexts, i.e., preferred fragment end motifs, which are defined as a few nucleotides at plasma DNA ends regardless of the site of origin within the genome. Detection of double-stranded plasma molecules carrying single-stranded protruding ends are termed jagged ends, which may be harnessed to assess the jaggedness across varying plasma DNA fragment sizes and their association with nucleosomal patterns. F Because ctDNA has a modal size profile shorter than that of the background cfDNA originating from non-cancerous cells, this fragment size feature can be used to enhance detection of tumor-associated alterations. Shorter fragments of cfDNA can be harvested either through specialized library preparation approaches that enrich for short cfDNA molecules, through in silico size selection approaches, or a combination of both. G Fragment size differences have also been shown to differentiate between mutations stemming from CHIP and those originating from the tumor. CHIP-associated mutations are associated with fragment size distributions of wildtype molecules (black distribution), whereas tumor-associated mutations typically reside on short cfDNA fragments (green distribution). H Using WGS data, global fragmentation patterns can be observed. By establishing coverage and size distribution references of cfDNA fragments in defined genomic windows in both healthy and cancer populations, it can be determined whether an individual’s cfDNA distribution is likely to be healthy (blue signal) or cancer-derived (red signal). By comparing genome-wide profiles between various tissues, these patterns may also be used for tissue deconvolution purposes. I Nucleosomes (purple circles) are shown in the form of heterochromatin or open chromatin regions along a length of DNA. Open chromatin consists of regulatory regions within the genome, such as enhancers, transcription factor binding sites (TFBS), promoters, and transcription start sites (TSSs), to which proteins may bind. These are highlighted in green and collectively represent DNase hypersensitivity sites (DHS). When a canonical nucleosome is supplanted, the underlying DNA is rendered accessible to nucleases and other protein factors