hg19, and the whole unfiltered NCBI-nt database
and were followed by aligned sequence counting
and annotation.
3 Results
3.1 Sensitivity of RNA and DNA Virus
Detection
Sequencing of four samples containing different
HSV-1 loads (10^1 ,10^2 ,10^3 , and 10^4 viral copies
per reaction) provided 135,846,381 (29.4–
37.3 million per sample) reads. The average
number of reads per sample was 33,961,595.
Similar numbers of reads were obtained for
HIV samples 130,7834,29 (25.5–38.8 million
per sample) and the average number of
sequences was 32,695,857. After trimming
there were 130,232,947 (28.2–32.2 million per
sample) reads for samples containing HSV-1 and
128,741,208 (25.1–38.0 million per sample)
reads for HIV-positive samples.
Viral sequences represented 0.001–0.620 %
of all reads in HIV-positive samples and
0.040–2.540 % of all reads in HSV-positive
samples. The results of sequence alignments are
shown in Table1 and Fig.1.
In the HIV RNA-positive CSF samples, the
highest number of HIV sequences were detected
in a sample containing 10^4 viral copies. In this
sample, 96.2 % of all viral reads were positively
aligned to the HIV genome (0.6 % of all reads,
coverage 97.0 % of HIV genome). In the
samples containing 10^3 and 10^2 viral copies,
88.5 % and 73.8 % of viral sequences, respec-
tively, aligned to the HIV genome constituted
0.022 % and 0.006 % of all reads (coverage
90.2 % and 30.0 % of HIV genome, respec-
tively). HIV sequences were not detected in the
sample with the lowest viral copy number (10^1 ).
The presence of HSV-1 DNA was detected
only in the two samples with the highest viral
titer of 10^3 and 10^4 and constituted 0.13 % and
25.90 % of all viral reads and 0.003 % and
0.028 % of all reads, respectively. The
HSV-specific reads covered 3.6 % and 13.0 %
of the viral genome at the concentrations of 10^3
and 10^4 copies/reaction, respectively. HSV-1
sequences were not detected in CSF containing
101 and 10^2 viral copies.3.2 Microbial ContaminationThe most abundant in both HIV and
HSV-positive samples were human sequences
(91.46–97.35 % of all sequences) followed by
bacterial sequences (1.04–4.14 % and
0.91–3.69 %); (Table 1). The results of
metagenomic analysis of two sterile water
samples (W1 and W2) and one control CSF sam-
ple (N) are shown in Tables2 and 3. The analysis
of the three negative controls was carried out in
three independent sequencing runs.
After trimming, there were 1,521,154 reads
for sample W1, 7,192,898 reads for sample W2,
and 40,340,940 reads for sample N (negative
CSF). Human DNA was common in the CSF
control sample (90.813 % of all reads), whereas
the sequences identified as ‘other’ (sequences
related to plants, plant viruses, and synthetic
constructs) dominated in both water samples
(31.419 % and 60.326 % of all reads). Human
sequences were less abundant in water (11.0 %
and 13.0 %) and the ‘other’ sequences
constituted only 3.116 % of all reads in
sample N.
Bacterial sequences represented 5.85 % and
21.40 % of all reads in water samples and 2.3o%
in the control sample N. The most frequent bac-
terial sequences corresponded toPsychrobacter
(sample W1),Acinobacter (sample W2), and
Corynebacterium (sample N). Other common
species included generaPseudomonas,Strepto-
coccus, Staphylococcus, Escherichia, Coryne-
bacterium, Bacillus, Propionibacterium and
Micrococcus(Table3).A total of 28 different
bacterial genera were common to all analyzed
samples. Viral, fungal, and protozoan reads
were less numerous and did not exceed 1 % in
any sample.56 I. Bukowska-Os ́ko et al.