CRAG:De novocharacterization of cell-free DNA fragmentation hotspots in plasma whole-genome sequencing

Abstract

The global variation of cell-free DNA fragmentation patterns is a promising biomarker for cancer diagnosis. However, the characterization of its hotspots and aberrations in early-stage cancer at the fine-scale is still poorly understood. Here, we developed an approach to de novo characterize genome-wide cell-free DNA fragmentation hotspots by integrating both fragment coverage and size from whole-genome sequencing. These hotspots are highly enriched in regulatory elements, such as promoters, and hematopoietic-specific enhancers. Surprisingly, half of the high-confident hotspots are still largely protected by the nucleosome and located near repeats, named inaccessible hotspots, which suggests the unknown origin of cell-free DNA fragmentation. In early-stage cancer, we observed the increases of fragmentation level at these inaccessible hotspots from microsatellite repeats and the decreases of fragmentation level at accessible hotspots near promoter regions, mostly with the silenced biological processes from peripheral immune cells and enriched in CTCF insulators. We identified the fragmentation hotspots from 298 cancer samples across 8 different cancer types (92% in stage I to III), 103 benign samples, and 247 healthy samples. The fine-scale fragmentation level at most variable hotspots showed cancer-specific fragmentation patterns across multiple cancer types and non-cancer controls. Therefore, with the fine-scale fragmentation signals alone in a machine learning model, we achieved 42% to 93% sensitivity at 100% specificity in different early-stage cancer. In cancer positive cases, we further localized cancer to a small number of anatomic sites with a median of 85% accuracy. The results here highlight the significance to characterize the fine-scale cell-free DNA fragmentation hotspot as a novel molecular marker for the screening of early-stage cancer that requires both high sensitivity and ultra-high specificity.

At the dawn: cell-free DNA fragmentomics and gene regulation

“…The concept of 'cfDNA fragmentomics' was first introduced by Ivanov et al in 2015 [32]. Since then, in addition to the study of the fragment coverages and sizes, several innovative computational and experimental approaches have been developed to comprehensively measure the cfDNAfragmentation patterns in plasma across different resolutions, including large-scale fragmentation patterns at megabase level (DELFI) [64], large-scale co-fragmentation patterns (FREE-C) [65], fragment coverage near transcription-start sites (TSS) [34], cfDNAaccessibility score near the transcription factor-binding sites (TFBS) [66], orientation-aware cfDNA fragmentation (OCF) [67], windowed protection score (WPS) [33], cfDNA-fragmentation hotspots [68], inference of DNA methylation from cfDNA-fragmentation patterns [69], the preferred-ended position of cfDNA [70,71], the end-motif frequency and motif-diversity score (MDS) [72], jagged end [73,74] and patterns outside the chromosomes [75][76][77][78]. Here, we will go through these state-of-art approaches applied at cfDNA fragmentation from a large-scale genomic bin to a single fragment.…”

“…Therefore, it is possible to evaluate the tissues-of-origin by cfDNA fragmentomics. DELFI, coverage near TFBS and cfDNA-fragmentation hotspots showed the potential to distinguish different cancer types in a supervised manner of machine learning but without providing the most relevant cell types contributed to cfDNA [64,66,68]. Preferred-end position, ended motif frequency and jagged end showed the potential for the estimation of the most relevant cell types but only demonstrated their correlations with the foetal sources in pregnant women, transplanted tissue source in organtransplantation patients and tumour sources in cancer patients [70][71][72]74].…”

FinaleDB: a browser and database of cell-free DNA fragmentation patterns

“…The size of cfDNA fragments differs according to their tissues-of-origin, including differences between fetal and maternal DNA, and between tumor and non-tumor derived DNA (Snyder et al, 2016). The cfDNA fragmentation patterns and their derived patterns from whole-genome sequencing (WGS), such as nucleosome positions, patterns near transcription start sites or transcription factor binding sites, ended position of cfDNA, fragmentation hotspots, co-fragmentation patterns, and large-scale fragmentation changes at mega-base level, offer extensive signals from the diseased tissues, as well as possible alterations from peripheral immune cell deaths, which can significantly increase the sensitivity for disease diagnosis (Snyder et al, 2016;Ulz et al, 2016;Jiang et al, 2018;Cristiano et al, 2019;Ulz et al, 2019;Sun et al, 2019;Liu et al, 2019;Zhou and Liu, 2020).…”