Recurrent sequencing artifacts at known variant positions: Illumina NovaSeq X + Agilent SureSelect Human All Exon V8

Laboratorio Aclimu (Buenos Aires, Argentina). Version 1, November 2026. License: CC BY 4.0. What this is Positions (GRCh38) where alternate reads appear recurrently in exome data from our clinical pipeline but are classified as sequencing artifacts. The sites evaluated are ClinVar pathogenic/likely pathogenic positions and canonical splice sites, inspected directly in the aligned reads (not called by GATK HaplotypeCaller). Several of these alleles are reported as pathogenic in ClinVar by other submitters; a low-fraction observation at these positions on this platform should not be reported without orthogonal confirmation. Data 24 exome samples: 20 clinical samples and 4 Genome in a Bottle reference runs (HG002, HG002 at 100x, HG003, HG004), all captured with Agilent SureSelect Human All Exon V8 and sequenced on Illumina NovaSeq X (2x150 bp). Alignment: BWA-MEM to GRCh38 (no-alt analysis set); duplicates marked. Reads with MAPQ >= 20. No patient-level data: only aggregate counts. Sites listed only if observed in >= 3 different samples. Artifact criterion At a site with alternate reads that was not called as a variant: all (or almost all) alternate reads on a single strand, with median alternate base quality < 10 and no support from base quality >= 20. Thirteen sites with alternate reads on both strands are kept and flagged (alt_reads_strand = both). Columns (TSV) chrom, pos, ref, alt, gene, transcript (MANE Select), hgvs_c, consequence; samples_with_artifact / samples_analyzed; giab_samples_with_artifact / giab_samples_analyzed; alt_reads_strand; median_alt_base_quality; median_vaf; median_depth; giab_hg002_v421_variant_at_position (whether the GIAB NIST v4.2.1 HG002 benchmark has any variant at the position; empty for chrX/Y/M); clinvar_variation_id, clinvar_classification, clinvar_stars (ClinVar release 2026-10-04); note (review notes, for example sites also called pathogenic/likely pathogenic with a low-quality filter by an independent clinical pipeline on the same data). hgvs_c is given on the MANE Select transcript, in transcript orientation. Files aclimu_novaseqx_v8_artifacts_v1.tsv: table. aclimu_novaseqx_v8_artifacts_v1.vcf: sites-only VCF (FILTER = ARTIFACT) with the same data in INFO. Limitations At ClinVar pathogenic/likely pathogenic positions only the ClinVar allele was screened, so other artifactual alleles at the same position may be missing; one such allele (MED13 c.4191+1G>C, 17 of 24 samples) was added after a cross-check with an independent pipeline. Platform- and chemistry-specific; derived from a modest number of samples from one laboratory. A site listed here can still carry a true variant in a given individual (for example a mosaic or a germline heterozygous variant with good-quality reads on both strands): the list flags positions that need orthogonal confirmation, it does not exclude variants. Contact [e-mail]