Transcriptomic Characterization of HPV+ and HPV- Head and Neck Cancer Patient-Derived Xenografts

RNA-Seq was used to characterize a large cohort of HNSCC patient-derived xenografts (PDXs) at the transcriptional level. Total RNA was extracted from lysates of whole PDX tumors using the Qiagen RNeasy kit according to the manufacturer’s instructions. The total RNA was enriched for polyA transcripts and sequenced at the QB3 Genomics center at the University of California, Berkeley, using the NovaSeq 6000 platform. After initial sequencing, the unaligned reads were derived from a combination of sources including human, mouse, viral, and ostensibly other sequences. Reads were classified to a probable source prior to mapping to the human reference and expression counting. Adapters were trimmed and reads failing QC removed using fastp (v0.20.0). Xenome (v1.0.0), a k-mer index based pseudoaligner, classified reads as "human," "mouse," "both," "neither," or "ambiguous". Reads identified as uniquely human were advanced to the human expression pipeline. rRNA matches were removed by aligning against a ribosomal reference with bwa. The remaining reads were mapped against an Ensembl GRCh38.90 reference, and expression counts were estimated using RSEM (v1.3.1). Library sizes were adjusted using the weighted trimmed mean of M-values method from the Bioconductor edgeR package and data were transformed to log counts per million for most analyses. Genes with median count less than 10 were excluded. We performed a special processing of the data to identify subtypes so that our data was on the same scale as data from the Cancer Genome Atlas (TCGA). Here we adjusted the library size using the upper quartile method from the edgeR package.