A Bioinformatics Workflow to Identify eccDNA Using ECCFP From Long-Read Nanopore Sequencing Data

Background

Extrachromosomal circular DNA (eccDNA) is a type of genetic material that exists independently of chromosomes and is widely found in eukaryotic organisms [1,2]. Recent studies have highlighted its significant role in various biological functions, particularly in cancer initiation, progression, and drug resistance [3–6]. In addition, eccDNA varies from tens to millions of base pairs (bp) in length, where molecules exceeding 10 kb are classified as ecDNA [7]. These large circular DNA molecules have been identified in over half of all human cancers, primarily using FISH and whole-genome sequencing (WGS) for validation [8,9]. Current experimental methods developed for the enrichment of small eccDNA, such as Circle-Seq [10] and 3SEP [11,12], typically identify eccDNA molecules smaller than 10 kb. These smaller eccDNA molecules have been demonstrated to regulate gene expression through various mechanisms [13].

Both Circle-Seq and 3SEP experimental methodologies utilize rolling circle amplification (RCA). During RCA, DNA polymerase continually traverses circular templates, producing concatemeric tandem copies (CTCs) [14]. Currently, most bioinformatics pipelines for eccDNA detection are based on overly strict selection criteria for CTCs [15] and often overlook the complexities introduced by rolling circle amplification (RCA) and random errors during the sequencing process, which might result in ghost sequences.

Recently, our group developed a novel bioinformatics pipeline, named EccDNA Caller based on Consecutive Full Pass (ECCFP) [16], for eccDNA detection from long-read Nanopore sequencing data. ECCFP utilizes all individual sequencing reads, including ghost and non-ghost sequences, to identify candidate eccDNA molecules. The pipeline then integrates these candidate eccDNAs and employs the Boyer–Moore majority vote algorithm to accurately determine eccDNA positions and gather information on circular consensus sequences. Compared to other current pipelines that identify small eccDNA molecules from long-read sequencing data using RCA amplification, ECCFP combines the strengths of these pipelines and is optimized specifically for RCA data, resulting in improved sensitivity, accuracy, and operational efficiency.

Briefly, the current signal data from Nanopore sequencing, typically stored in FAST5 or POD5 formats, undergo base calling through tools such as Guppy or Dorado. This computational conversion transforms the current signal data into nucleotide sequence data, ultimately producing demultiplexed FASTQ files for downstream genomic analyses. Upon FASTQ acquisition, initial quality assessment is performed using NanoPlot. Adapter and barcode sequences are then removed with Porechop, followed by post-trimming quality re-evaluation of trimmed reads using NanoPlot. Trimmed reads require alignment to a reference genome for sequence mapping. Minimap2 serves as the optimal aligner for long-read data due to its efficient handling of error-prone sequences. Subsequent eccDNA detection is conducted using ECCFP, which identifies circular DNA through structural signature analysis of aligned reads. The entire process from Nanopore sequencing data acquisition to eccDNA identification can be referenced to in the Graphical Overview for clearer understanding. This protocol details the core bioinformatic workflow for eccDNA detection, encompassing quality control, adapter and barcode trimming, reference genome alignment, and circular DNA identification. As a computational pipeline, it exclusively addresses analytical methodologies while explicitly excluding wet laboratory procedures. All sequencing data used are derived from public repositories, including Project PRJCA040952 and PRJCA052047 deposited by our group at NGDC alongside datasets from published studies.

This protocol relies exclusively on open-source software tools that are readily accessible. Furthermore, we provide comprehensively documented analytical code designed to guide users through the complete workflow from FASTQ processing to eccDNA identification. This makes the protocol accessible even for researchers without prior bioinformatics experience.