A Tumor-Only Whole Genome Sequencing Pipeline for the Development of Molecular Pathology Assays for AML

• Published in: Blood 2024-11, Vol.144 (Supplement 1), p.7479-7479
• Main Authors: Velazquez-Muriel, Javier, Tyndale, Selene, D'Angelo, Julian, Sanchez, Paulina, Shah, Presha, Huang, Ying, Miller, Jeffrey E, Niemeijer, Meindert
• Format: Article
• Language: English
• ISSN: 0006-4971; 1528-0020
• DOI: 10.1182/blood-2024-197974

Introduction The discovery of the genomic variants that are part of Acute Myeloid Leukemia (AML) is of great importance for its diagnosis, clinical subtype classifications, treatment, and prognosis. The current state of the art for variant detection utilizes next generation sequencing techniques, typically sequencing only selected regions of the genome involving the most-frequently affected genes (targeted gene panels), or all coding regions. With decreasing costs, the use of whole genome sequencing (WGS) is becoming attractive, as it offers the advantage of uncovering variants outside gene panels or coding regions, a feature of key importance for AML, where structural variants, and in particular gene fusions, are known to play a significant role. We present a tumor-only, somatic WGS pipeline for the development of molecular pathology assays for AML, and potentially, other diseases. Methods We built a cloud-based WGS pipeline consisting of 1) sequencing WGS libraries prepared from AML samples using the Illumina DNA Prep workflow on an Illumina NovaSeq 6000 sequencer and S4 flow cell, generating 72-108X coverage per sample, 2) demultiplexing on local hardware to generate sample FASTQ read files, and 3) : read-mapping to the reference genome, quality metrics gathering, deduplication of reads, base quality recalibration, variant calling (single nucleotide variants and small indels), and variant filtering. We compared the results of our WGS pipeline to those obtained with Invivoscribe's gene panel for 10 clinical samples, 2 AML cell lines (OCI-AML3, MV4-11), and a negative control (HG002). We also ran our WGS pipeline with a reference sample (Seracare Myeloid Mutation DNA mix). In all cases, we looked for detection of the Pathogenic/Likely Pathogenic (P/LP) variants and their variant allele fraction (VAF). Results Our WGS pipeline was able to from the reference sample. Compared to MyAML, the WGS pipeline achieved 100% (44/44) detection of the P/LP variants present at for the other samples. Of the 15 P/LP variants with VAF < 5% reported by MyAML, only 1 was detected by the WGS pipeline. The average absolute difference for the VAF value between MyAML and the WGS pipeline was 10% or less for . Our cloud WGS pipeline can process an entire sample from NovaSeq run in 4h or less. Conclusions Our WGS pipeline for analyzing WGS data from AML samples can be an effective substitute of the current MyAML gene panel for variants detection with VAF > 5%, . We are working on