ECCFP: A consecutive full pass-based bioinformatic analysis for eccDNA identification from long-read sequencing data

Emerging studies focused on extrachromosomal circular DNA (eccDNA) due to its various functions and complex mechanisms. The main technique for eccDNA detection is using high-throughput sequencing followed by bioinformatics analysis, but with shortcomings such as low accuracy and efficiency. To address these limitations, we developed ECCFP, a bioinformatics pipeline that detects eccDNAs from long-read sequencing data with rolling circle amplification. ECCFP retains all consecutive full passes from individual reads, including those from ghost sequences, to obtain the candidate eccDNAs. Subsequently, ECCFP employs the Boyer-Moore majority voting algorithm to consolidate the multiple candidate eccDNAs to generate an accurate unique eccDNA and circular consensus sequence. By using multiple simulated datasets and real sequencing datasets with spike-in eccDNAs or publicly available, we conducted comprehensive evaluations of ECCFP and compared it with three currently published pipelines that have demonstrated superior performance. The results indicate that ECCFP outperforms other pipelines in sensitivity, accuracy, and runtime efficiency. Furthermore, outward PCR amplification followed by Sanger sequencing validation confirmed that ECCFP detects more accurate eccDNA junction position than other pipelines. Conclusively, ECCFP is an efficient and accurate bioinformatic pipeline for eccDNA detection from long-read sequencing data, which might provide a robust tool for eccDNA characterization in various biological samples.