High-fidelity whole genome sequencing from low-input FFPE samples: Enabling accurate tumor-informed MRD assay design through superior variant detection and uniform coverage

Introduction: Tumor-informed Minimal Residual Disease (MRD) monitoring relies on the precise identification of patient-specific somatic variants from primary tumor tissue. However, these samples are frequently archived as Formalin-Fixed Paraffin-Embedded (FFPE) blocks, where DNA damage, fragmentation, and chemical modifications (e.g., cytosine deamination) compromise sequencing quality. Inaccurate baseline profiling can lead to the selection of artifacts as tracking targets or the omission of critical sub-clonal mutations, ultimately reducing MRD assay sensitivity. Here, we present a high-fidelity Whole Genome Sequencing (WGS) workflow optimized for low-input FFPE samples that minimizes amplification bias and artifacts, ensuring the generation of reliable genomic maps necessary for robust MRD target selection. Methods: Genomic DNA was extracted from FFPE-processed NA12878 cells (FF12878) using the Covaris truXTRAC® FFPE SMART Kit with Adaptive Focused Acoustics (AFA®) technology to ensure active paraffin removal and rehydration. WGS libraries were constructed using the truCOVER® WGS Library Prep Kit with Amplification. The workflow was evaluated using DNA inputs ranging from 1 ng to 50 ng. Library performance was benchmarked against high-quality genomic DNA (NA12878) inputs (0.1 ng - 50 ng). Libraries were sequenced on an Illumina NovaSeq™ X Plus. Data was analyzed for coverage uniformity, duplication rates, and variant calling accuracy (F1 scores) for SNPs and INDELs using the GIAB truth set. Results: The optimized workflow demonstrated exceptional sensitivity, generating high-complexity libraries from as little as 1 ng of FFPE DNA. Critical for MRD design, the method exhibited minimized GC bias, maintaining normalized coverage centered around 1.0 across 504 clinically relevant cancer genes (TSO500 panel), irrespective of input amount. This uniformity ensures that potential tracking mutations in difficult-to-sequence regions are not missed. Variant calling analysis revealed that the workflow effectively overcomes FFPE-induced damage; F1 scores for SNPs and INDELs in FFPE samples (5 ng) were comparable to those of high-quality gDNA (>0.91 for INDELs and >0.97 for SNPs). Furthermore, duplication rates were significantly controlled (~19% for 5 ng FFPE), maximizing the unique read depth required to validate low-frequency truncal mutations essential for longitudinal tracking. Conclusions: We have validated a robust WGS workflow that unlocks archival FFPE tissues for high-fidelity genomic profiling, overcoming traditional barriers of low input and DNA damage. By delivering superior coverage uniformity and high variant calling accuracy, this solution provides the precise genomic baseline required for designing highly specific tumor-informed MRD assays