I am investigating a suspected duplication of exon 8 in the OPA1 gene, which was initially identified and confirmed via Multiplex Ligation-dependent Probe Amplification (MLPA). To characterize the structural variant (SV) at base-pair resolution and resolve its genomic architecture, we performed whole-genome sequencing (WGS) using Oxford Nanopore Technologies (ONT) long-read sequencing (LRS).

Bioinformatics Pipeline & Current Results:

• Basecalling & Alignment: Raw data was basecalled using Dorado to generate a coordinate-sorted BAM file mapped against the human reference genome .
• SV Calling: I ran cuteSV and Sniffles to detect structural variants using default paramaters, but neither caller flagged the exon 8 duplication or any explicit SV at the OPA1 locus.
• Depth & Variant Allele Frequency (VAF) Analysis: Interestingly, when analyzing the single nucleotide polymorphisms (SNPs) across the region, a Manhattan-style plot tracking sequencing depth and allele frequencies (AF) revealed a distinct, elevated coverage block extending from exon 7 to exon 23, suggesting a much larger duplication event than the single-exon signal caught by MLPA.

I am trying to determine whether this large increase in exon 7-23 coverage represents a true positive or a complex artifact. I am currently considering performing a local alignment (e.g., BLASTN) of the original MLPA probe sequences directly against the phased/unphased LRS reads to verify their binding architectures. I want to rule out whether this is a localized tandem duplication or a complex insertion into a different chromosome (an interchromosomal translocation).

Questions for the Community:

1. Is executing a local sequence alignment of MLPA probes against the ONT reads a robust strategy to confirm the duplication breakpoint, or will the raw error rate of ONT reads confound local short-probe alignments?
2. Why might top-tier long-read SV callers like cuteSV or Sniffles fail to identify a duplication that shows a clear copy-number/depth signature across multiple exons? Are there specific parameter tuning flags (e.g., minimum support reads, max cluster size) recommended for mid-sized exonic duplications in ONT data?
3. What are the best alternative in silico approaches or specialized long-read copy number variant (CNV) tools (e.g., CNVpytor, Mosdepth for binned depth analysis) to validate this finding and resolve whether it's tandem or ectopic?