This month, our team at Golden Helix is proud to highlight a series of standout customer publications demonstrating the real-world impact of next-generation sequencing and variant analysis. Featured articles from June 2025 showcase the power of the VarSeq platform in supporting high-quality research across hereditary and somatic disease.
From integrating AI to enhance variant detection in colorectal cancer, to reclassifying variants in a large Danish cohort with suspected hereditary breast and ovarian cancer, and investigating somatic mosaicism in a rare case of hereditary hemorrhagic telangiectasia, each study reflects the dedication of our customers to advancing genomic medicine and improving patient care.
Title: Integrating next-generation sequencing and artificial intelligence for the identification and validation of pathogenic variants in colorectal cancer
Title: Integrating next-generation sequencing and artificial intelligence for the identification and validation of pathogenic variants in colorectal cancer
Background: Colorectal cancer (CRC) arises from a combination of genetic and environmental factors, making early detection of pathogenic variants essential. Next-generation sequencing enables comprehensive multigene analysis for both hereditary and sporadic CRC cases.
Objectives: To identify pathogenic and likely pathogenic germline variants in Colombian CRC patients and evaluate the effectiveness of AI-based tools in variant classification and functional validation.
Subjects and Methods: Germline variants were analyzed in 100 unselected Colombian CRC patients using ACMG/AMP guidelines and the BoostDM AI model to identify oncodriver mutations. Results were compared with AlphaMissense predictions, and a minigene assay was used to functionally validate intronic mutations.
Results
- Pathogenic or likely pathogenic (P/LP) variants were identified in 12% of patients using ACMG/AMP criteria.
- BoostDM identified oncodriver variants in 65% of cases, demonstrating the value of AI integration.
- Average AUC scores comparing BoostDM to AlphaMissense were 0.788 overall and 0.803 for panel genes.
- Functional validation via minigene assay revealed aberrant transcripts linked to CRC etiology.
Conclusions: The study demonstrated that integrating NGS with artificial intelligence enhances the detection of pathogenic germline variants in unselected Colombian CRC patients. Functional validation of intronic variants further supports the need for a multifaceted approach to understanding the genetic complexity of colorectal cancer.
“Variant call format (VCF) files were analyzed using the software VarSeq® (Golden Helix, v 2.3.0). We incorporated the following database annotations: ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), Ensembl (https://www.ensembl.org/index.html), RefSeq (https://www.ncbi.nlm.nih.gov/refseq/), dbNSFP Functional predictions, dbSNP, REVEL, OMIM Phenotype Ontology (https://www.omim.org/), UniProt Variants (https://www.uniprot.org/) and gnomAD v2.1 (https://gnomad.broadinstitute.org/).”
Title: Classification of Gene Variants in a Danish Population with Suspected Predisposition to Hereditary Breast and/or Ovarian Cancer
Background: This study analyzed genetic data from 5,923 Danish patients with suspected hereditary breast and/or ovarian cancer, reclassifying 22.8% of previously uncertain gene variants and finding that 10.6% carried likely pathogenic or pathogenic variants. The high number of variants of unknown significance highlights the ongoing challenges in clinical genetic counseling.
Objectives: To classify gene variants in a large cohort of Danish patients with suspected hereditary breast and/or ovarian cancer. It focuses on re-evaluating variants of unknown significance (VUSs) using association and splice analyses to support clinical interpretation.
Subjects and Methods: Patients were referred from three major Danish hospitals and analyzed using a 13-gene NGS panel between 2012–2022. Variant classification followed ACMG and gene-specific guidelines, incorporating annotation, CNV calling, population frequency comparison, and splice analysis. Statistical association testing was performed using allele frequencies and validated tools, including VarSeq and R, with additional splice impact assessed via RNA sequencing in selected cases.
Results:
- Out of 6,824 initial samples, 5,923 met quality criteria and were included in the analysis.
- Likely pathogenic or pathogenic (LP/P) variants were detected in 10.6% of patients, primarily in BRCA1 and BRCA2.
- Variants of unknown significance (VUSs) were identified in 27.1% of patients, most frequently in BARD1 and ATM.
- While association and splice analyses added further context, most VUSs remained unclassified—highlighting the ongoing challenges in clinical interpretation.
“VCF files were imported to VarSeq (Golden Helix, Bozeman, MT, USA) for annotation and filtering. Variants were annotated using RefSeq transcripts and filtered by sequencing ontology and the gene-specific BA1 (benign stand-alone criterion in ACMG guidelines) population frequency.
VarSeq CNV caller (Golden Helix, Bozeman, MT, USA) was used to call copy number variants, using whole exons as the target region.”
Munch, A.K.; Feldner, E.S.; Bækgaard, C.H.; Larsen, M.B.; Slemming-Adamsen, N.; Boonen, D.S.; Møller, N.B.; Pedersen, I.S.; Hansen, T.V.O.; Terkelsen, T.; et al. Classification of Gene Variants in a Danish Population with Suspected Predisposition to Hereditary Breast and/or Ovarian Cancer. Cancers 2025, 17, 1819. https://doi.org/10.3390/cancers17111819
Title: Multiple Lesion-Specific Somatic Mutations and Bi-Allelic Loss of ACVRL1 in a single patient with Hereditary Haemorrhagic Telangiectasia
Background: Hereditary Haemorrhagic Telangiectasia (HHT) is an inherited vascular disorder typically caused by pathogenic variants in ENG, ACVRL1, or SMAD4, leading to telangiectasias and arteriovenous malformations (AVMs). Recent findings suggest that somatic mutations in the remaining functional allele may contribute to lesion development, prompting further investigation into the underlying mechanisms.
Objectives: To investigate the role of somatic mosaicism in HHT lesion development by analyzing multiple vascular lesions from a single patient with a germline deletion of the ACVRL1 gene. It seeks to identify lesion-specific somatic variants that may contribute to forming AVMs and telangiectasias.
Subjects and Methods: Deep exome sequencing was performed on DNA from multiple fresh tissue biopsies from the same HHT patient: six hepatic AVM samples, two macroscopic normal hepatic control samples, and three mucocutaneous telangiectasia biopsies.
- Somatic mosaic lesion-specific ACVRL1 variants were identified in four hepatic AVM samples and one skin telangiectasia.
- Two different somatic variants were detected in multiple lesions from the same liver.
- A third lesion-specific somatic variant was identified in one skin telangiectasia sample.
Results:
“Sample preparation was performed using 400 ng DNA after Covaris mechanical fragmentation. Library preparation was conducted following the Twist Exome 2.0 hybridization protocol (Twist Bioscience, Inc.). Validated libraries were pooled and sequenced in paired-end mode 2×150 bp the on Illumina NovaSeq 6000 platform (Illumina, Inc.). The sequencing quality criteria obliged a minimum coverage of 20x in 95% of the coding regions. The mean coverage was around 1200X.”Haahr, P., Hao, Q., Brusgaard, K., Larsen, M., Lange, B., Fialla, A., … & Torring, P. (2025). Multiple lesion-specific somatic mutations and bi-allelic loss of acvrl1 in a single patient with hereditary haemorrhagic telangiectasia.. https://doi.org/10.21203/rs.3.rs-6438890/v1