Examining intra-host genetic variation of RSV by short read high-throughput sequencing

Abstract

Every viral infection entails an evolving population of viral genomes. High-throughput sequencing technologies can be used to characterize such populations, but to date there are few published examples of such work. In addition, mixed sequencing data are sometimes used to infer properties of infecting genomes without discriminating between genome-derived reads and reads from the much more abundant, in the case of a typical active viral infection, transcripts. Here we apply capture probe-based short read high-throughput sequencing to nasal wash samples taken from a previously described group of adult hematopoietic cell transplant (HCT) recipients naturally infected with respiratory syncytial virus (RSV). We separately analyzed reads from genomes and transcripts for the levels and distribution of genetic variation by calculating per position Shannon entropies. Our analysis reveals a low level of genetic variation within the RSV infections analyzed here, but with interesting differences between genomes and transcripts in 1) average per sample Shannon entropies; 2) the genomic distribution of variation 'hotspots'; and 3) the genomic distribution of hotspots encoding alternative amino acids. In all, our results suggest the importance of separately analyzing reads from genomes and transcripts when interpreting high-throughput sequencing data for insight into intra-host viral genome replication, expression, and evolution.

Fig 1.. Transcripts exceed genomes by ~100x and reads derived from both show fairly uniform coverage of reference RSV genomes.

The interquartile range of per position sequencing depth from genome- (in red) and transcript-derived read sets (in blue) is plotted along the RSV genome for both RSV/A/ON (top plot) and RSV/B/BA references (bottom plot). Darker lines represent the upper bounds of Q3 and Q1.

Fig 2.. Genomes are more variable than transcripts but both contain highly variable positions located across the RSV genome in different samples.

The interquartile range of per position Shannon entropy (H) from genome- (in red) and transcript-derived read sets (in blue) is plotted along the RSV genome for both RSV/A/ON (top plot) and RSV/B/BA references (bottom plot).

Fig 3.. Per sample average or bulk Shannon entropies of genomes and transcripts differ in magnitude and dynamics.

Plots of per sample average Shannon entropy ($H$) vs. day of sample acquisition. All per position Shannon entropies for single samples were averaged for genome- (left plot) and transcript-derived read sets (right plot). RSV/A/ON data in blue; RSV/B/BA data in red; data from subjects who shed RSV for < 14 days in closed circular points; data from subjects who shed RSV for ≥ 14 days in closed triangular points.

Fig 4.. ‘Hotspots’ of variation are more numerous and densely distributed across the RSV genome in genome-derived reads than transcripts.

Hotspots or single positions showing exceptionally high variation (Shannon entropy (H) ≥ 0.3) are plotted across the genome for both genome- (left plots) and transcript-derived read sets (right plots) from RSV/A/ON (top plots) and RSV/B/BA (bottom plots) infections. Each plot contains data for the number of samples indicated in parentheses. Each line in each plot shows the genomic distribution of hotspots for a single sample. Hotspots are colored according to their position within either 1) different noncoding sequences or 2) the 10 coding sequences of RSV.

Fig 5.. ‘Hotspots’ of functional variation are more numerous and densely distributed across the RSV genome in genome-derived reads than transcripts.

Hotspots or single positions showing exceptionally high variation (Shannon entropy $\left(H\right)\ge 0.3$) and encoding alternative amino acids are plotted across the genome for both genome- (left plots) and transcript-derived read sets (right plots) from RSV/A/ON (top plots) and RSV/B/BA (bottom plots) infections. Each plot contains data for the number of samples indicated in parentheses. Each line in each plot shows the genomic distribution of hotspots encoding one or more alternative amino acids for a single sample. Hotspots are colored according to their position within the 10 coding sequences of RSV.

Fig 6.. Amino acid (AA) Shannon entropies are higher across the RSV genome in genome-derived reads than transcripts.

The interquartile range (IQR) of per position amino acid Shannon entropy ($H$) from genome- (left plots) and transcript-derived read sets (right plots) is plotted along the RSV genome for both RSV/A/ON (top plots) and RSV/B/BA references (bottom plots). Bars showing the IQR of AA $H$ are colored according to their position within the 10 coding sequences of RSV.