Mining for Viral Expression in Human Cancer

Human papillomavirus (HPV) is known to be linked to cervical cancers in humans, and luckily HPV vaccines have become fairly common. The question of what viruses in general play a role in cancers is a fascinating one with much more work to be done to answer questions. Next generation sequencing technologies and bioinformatics technologies enable us today to explore this question from a new and powerful perspective computational. Rich data sets such as The Cancer Genome Atlas (TCGA) database contain thousands of digital samples of sequenced tumor and normal tissues. During my work with the National Cancer Institute, we used this data set to explore the prevalence of viral expression in different cancer and normal tissue.

Cervical cancer is known to be linked to HPV, which contains oncogenes that have been shown to induce cellular transformations that can lead to malignancies. So we expected at least some expression of HPV genes in these cancers. Brain cancers (glioblastoma) on the other hand provide a great control sample, as viruses are not be able to cross the blood-brain barrier and thus no expression would be expected.

We compared the expression of HPV in cervical, brain, and ovarian cancer and normal tissue RNAseq data using sequence alignment software BowTie2, developed by former colleagues of mine at the University of Maryland. The way this works, is that one gives the software the sample sequence and a reference sequence – in this case the reference is the HPV DNA sequence – and the alignment software can be used to search for matches in the sample sequence to the reference. Reference sequences for all kinds of viruses and organisms can be downloaded from the National Center for Biotechnology Information. We ran the sequence alignment and analysis on NIH’s supercomputing cluster.

The results are published in this paper.¬†Almost all (98%) of the cervical tumor samples produced alignments to oncogenes of at least one of the 12 high-risk HPV types. In contrast, none of the brain tumor samples produced alignments. A very low percentage of ovarian cancers expressed the oncogenes of HPV. Various questions are still open about what other cancers in males and females might have a causal link to HPV. For me, although I had heard about the HPV and cervical cancer link before from experts’ vaccine recommendations, seeing the staggering cervical numbers first hand left a permanent impression on me regarding their importance.