The median per-base coverage of the input DNA sequence reads was 16-fold, and for the genomic DNA sequence reads was 8-fold, giving a genome-wide ratio of 2. We then ranked all windows from least- to most-covered by input sequence reads, normalized by the genomic read counts.
No evidence supporting a weaker expression of these genes was detected (median reads per base: histones 3.1, all other genes 2.9).
In total, 18,894 RNA-derived contigs were assembled (comprising 22.1 Mbp of transcriptome sequence) that were greater than 200 bp in length (mean = 1170 bp, Figure 1 and Additional file 2), with a median coverage per base (CPB) per contig of 37×, ranging from 8× (minimum coverage cut-off for assembly) to 5,262× (Additional file 1 Figure S2).
Since median per-visit adherence is based on data at all visits, participants who attend more visits will contribute more information.
Within the diploid index, genes had a median intergenomic SNP per base rate of 2.2% (range, 0 16.1%).
Therefore they were trimmed up to an optimal length sufficient to increase the median QS value per base to at least 25.
Boxplot distributions of quality scores (QS) for 454 sequencing showed that the median QS value per base was above 20, up to the first 350 nucleotides in each read.
In particular, upon the observation of quality metrics generated by the FastQC program, OST-78 and OST-83 reverse-reads were trimmed starting from the180th and the 209th (OST-83) nucleotide of the sequence, respectively, in order to maintain a median QS value per base ≥ 25, overall the reads.
Since a drop in median quality score (QS) per base was observed only within the 3′end of MiSeq reverse-reads in each sample, these datasets were trimmed using custom python scripts.
The median (IQR) deep sequencing coverage per base was 3938 3130 6945945) sequences at baseline and 3965 3688 7091091) sequences at virological failure.
