Virasign reports NOGR as an extra per-virus evidence metric: it counts how many distinct, non-overlapping regions of the reference genome are supported by at least one mapped read, and the total number of reference bases spanned by those regions. This follows an mNGS reporting idea used before that can be very valuable for interpretation: requiring a minimum number of non-overlapping viral reads/regions to support a detection (e.g. Nature Communications s41467-024-51470-y, which uses a threshold of ≥3 non-overlapping viral reads/contigs as a positive criterion).
Coverage breadth can be low for true positives (degraded samples, low viral load), but it can also be low for technical artifacts such as amplicon contamination, where many reads stack on the same small viral genomic region. NOGR helps distinguish real signals from false contamination.
Virasign looks at the virus BAM file ({accession}.bam) and does this:
- Step 1: For each mapped read, record where it maps on the virus genome (start and end coordinates).
- Step 2: Count how many reads map to different parts of the genome (reads that do not overlap each other on the genome).
This gives two values:
nogr_regions: how many separate (non-overlapping) genomic regions are supportednogr_bases: how many virus genome bases are covered by those separate regions (so it can never be larger than the genome length)