Evaluation of reference sample type for somatic variant calling in myeloid cancers

High-throughput sequencing is transforming cancer diagnostics, research, and clinical decision-making, yet accurate identification of disease-driving somatic variants often requires comparing tumor and normal DNA. This study aimed to identify the optimal normal reference sample for whole exome sequencing (WES) in myeloid-derived cancers. T-cells, skin biopsies, saliva, and a tumor-only analysis were evaluated by comparing variant calls from tumor-reference pair to those from tumor-fibroblast analyses. T-cells demonstrated the highest sensitivity (0.91-1.00) and lowest false positive rate, establishing them as a reliable reference option. Skin biopsies and saliva yielded inconsistent results, likely due to leucocytic contamination. While tumor-only analysis offered high sensitivity, it generated numerous false positives; however, a selective filtration strategy provides a feasible solution by omitting the need for a normal reference sample. These findings suggest T-cells are the optimal balance of accuracy and practicality for identifying somatic variants in myeloid cancers, while tumor-only analysis presents a viable alternative with careful filtering. This improved workflow supports efficient and reliable genomic profiling for personalized treatment decisions. ClinicalTrials.gov, TRN: NCT05695638, Registration date: 10 January 2023.