Early detection of lung cancer is important for improving patient survival rates. Liquid biopsy using whole-genome sequencing of cell-free DNA (cfDNA) offers a promising avenue for lung cancer screening, providing a potential alternative or complementary approach to current screening modalities. Here, we aimed to develop and validate an approach by integrating fragment and genomic features of cfDNA to enhance lung cancer detection accuracy across diverse populations. Deep learning–based classifiers were trained using comprehensive cfDNA fragmentomic features from participants in multi-institutional studies, including a Korean discovery dataset (218 patients with lung cancer and 2,559 controls), a Korean validation dataset (111 patients with lung cancer and 1,136 controls), and an independent Caucasian validation cohort (50 patients with lung cancer and 50 controls). In the discovery dataset, classifiers using fragment end motif by size, a feature that captures both fragment end motif and size profiles, outperformed standalone fragment end motif and fragment size classifiers, achieving an area under the curve (AUC) of 0.917. The ensemble classifier integrating fragment end motif by size and genomic coverage achieved an improved performance, with an AUC of 0.937. This performance extended to the Korean validation dataset and demonstrated ethnic generalizability in the Caucasian validation cohort. Overall, the development of a deep learning–based classifier integrating cfDNA fragmentomic and genomic features in this study highlights the potential for accurate lung cancer detection across diverse populations. Significance: Evaluating fragment-based features and genomic coverage in cell-free DNA offers an accurate lung cancer screening method, promising improvements in early cancer detection and addressing challenges associated with current screening methods.

Jung KY, et al. J Korean Med Sci. 2018 May;33(29):e191. https://doi.org/10.3346/jkms.2018.33.e191

Genomic instability of circulating tumor DNA (ctDNA) as a prognostic biomarker has not been evaluated in pancreatic cancer. We investigated the role of the genomic instability index of ctDNA in pancreatic ductal adenocarcinoma (PDAC). We prospectively enrolled 315 patients newly diagnosed with resectable (n = 110), locally advanced (n = 78), and metastatic (n = 127) PDAC from March 2015 through January 2020. Low-depth whole-genome cell-free DNA sequencing identified genome-wide copy number alterations using instability score (I-score) to reflect genome-wide instability. Plasma cell-free and matched tumor tissue DNA from 15 patients with resectable pancreatic cancer was sequenced to assess the concordance of chromosomal copy number alteration profiles. Associations of I-score with clinical factors or survival were assessed. Seventy-six patients had high genomic instability with I-score > 7.3 in pre-treatment ctDNA; proportions of high I-score were 5.5%, 5.1%, and 52% in resectable, locally advanced, and metastatic stages, respectively. Correlation coefficients between Z-scores of plasma and tissue DNA at segment resolution were high (r² = 0.82). Univariable analysis showed the association of I-score with progression-free survival in each stage. Multivariable analyses demonstrated that clinical stage-adjusted I-scores were significant factors for progression-free and overall survival. In these patients, ctDNA genomic I-scores provided prognostic information relevant to progression-free survival in each clinical stage.

Next-generation sequencing (NGS) technology has become a trend in the genomics research area. There are many software programs and automated pipelines to analyze NGS data, which can ease the pain for traditional scientists who are not familiar with computer programming. However, downstream analyses, such as finding differentially expressed genes or visualizing linkage disequilibrium maps and genome-wide association study (GWAS) data, still remain a challenge. Here, we introduce a dockerized web application written in R using the Shiny platform to visualize pre-analyzed RNA sequencing and GWAS data. In addition, we have integrated a genome browser based on the JBrowse platform and an automated intermediate parsing process required for custom track construction, so that users can easily build and navigate their personal genome tracks with in-house datasets. This application will help scientists perform series of downstream analyses and obtain a more integrative understanding about various types of genomic data by interactively visualizing them with customizable options.

Experimental Batch Effect on FEMS Feature Before and After Bias Correction.

Supplementary Table S5. The cancer prediction scores calculated by each classifier in the Korean, and Caucasian datset.