Resolution of germline hereditary cancer structural variants using nanopore sequencing

My Linh introduced the Personalised OncoGenomics program (POG) at her institute, an open clinical trial investigating advanced cancers using very comprehensive multi-omics analyses. Whole genome and transcriptome analyses are being performed to identify a wide range of variant classes. Currently, the germline sequencing data is being reviewed, but they have looked at somatic structural variants and validated these using the Oxford Nanopore MinION.

Structural variants (SVs) are difficult to resolve using short-read sequencing. My Linh described how she has applied long-read nanopore sequencing on the MinION to validate and resolve 4 germline SVs present in hereditary cancer genes. Initially, the presence of an SV had been detected from short read alignments. The SVs validated were: a partial ATM deletion; an NTHL1-TSC2-TRAF7 complex rearrangement; an IFT140-TSC2 inversion; and an UIMC1-NSD1 rearrangement. These were selected for long-read sequencing validation as they were either complex, the clinical phenotype did not match that expected, or they could not be resolved using short reads, and there was a potential clinical action related to the findings.

From 12.6 Gb MinION sequencing data, the ATM partial deletion was detected from a single 13 kb read across the breakpoint. My Linh explained how this was identified computationally using the alignment tool NGMLR (minimap2 was unsuccessful as it clipped the read); the deletion was consistent when investigated using long-range PCR. My Linh highlighted the benefits of nanopore long-read sequencing in its ability to sequence across repetitive regions, resolving the exact size of the deletion. She stated that nanopore sequencing is a good alternative to long-range PCR which is more expensive, and standard PCR was not unable to identify it.

The second SV that My Linh highlighted was a complex inversion-deletion across an 187 kb region containing genes NTHL1, SC2, and TRAF7, present in an individual with a tuberous sclerosis phenotype. Short-read sequencing had been unable to detect this complex rearrangement, but it was identified by 4 long nanopore reads traversing the locus; the presence of 1 wildtype read across the region suggested that the rearrangement was present on only one allele. The individual's tumour was characterised by a significantly high presence of the COSMIC database mutational signature 30; this signature is highly associated with biallelic germline NTHL1 aberrations, and so it was assumed that this patient's was also biallelic. However, with long read sequences they identified that in fact only one allele was disrupted and they had been biased in their assumptions due to their knowledge of signature 30.

My Linh next discussed the case of 3 patients with the same IFT140-TSC2 inversion. A clinical phenotype of tuberous sclerosis tumours was expected, but was in fact only possibly associated with 1 case; 2 of the 3 tumours were instead associated with non small cell lung carcinoma and colon adenocarcinoma. The suspected IFT140-TSC2large inversion in one patient was found to be a small (132 bp) inverted-duplicated TSC2 intronic sequence inserted into an intron of IFT140; this insertion did not disrupt TSC2, and so what was suspected to be a pathogenic tumour was likely to have been benign.

The final complex SV described was an individual with a complexinversion-duplication rearrangement across a 207 kb region, containing two genes UIMC1 and NSD1. The two breakpoints were identified on one allele by 7 long nanopore reads. My Linh explained how these results determined that the SV of previously uncertain significance was likely to have a benign effect, and how this rearrangement was much easier to map using long reads.

My Linh summarised her talk by highlighting how long-read nanopore sequencing was required for the accurate resolution of what had been inaccessible SVs - all 12 breakpoints for the 4 SVs described were present only in the nanopore sequencing data. She suggested that whole-genome sequencing with long nanopore reads should be considered for the precise detection and validation of clinically relevant germline SVs. The "potential I see for long reads is comprehensive assessment of all types of structural variants."

Her future plans include PromethION sequencing for a pilot study on male breast cancer genes, which will require precise breakpoint characterisation.

Authors: My Linh Thibodeau