The performance of the target capture panel can be evaluated using various metrics and below are few examples for defining the performance:
1. Coverage
There are two types of coverage: depth of coverage and breadth of coverage.
Depth of coverage (usually abbreviated as ‘depth’) refers to the average number of reads that aligned to a region of interest. It is calculated by dividing the total number of reads that align to the target region by the size of the aligned region. For example, if a specific genomic position is covered by 1000 reads and the size of the aligned region is 100 base pairs, the depth for that region is 10X.
Breadth of coverage (usually abbreviated as ‘coverage’) refers to the proportion of a genome or target region that is covered by sequencing reads to a certain depth. It is calculated by dividing the number of bases with a specified depth of coverage by the total number of bases in the target region or genome. For example, if a target region of 1000 base pairs is covered by at least 20 reads for 900 base pairs, the 20X coverage for that region is 90%.
High breadth and depth of coverage are generally desirable for accurate variant detection and downstream analysis.
2. On-target ratio
On-target ratio refers to the proportion of sequencing reads that map to the target regions. It is calculated by dividing the number of reads that align to the target regions by the total number of sequencing reads generated in the experiment. For example, if a targeted gene panel generates 10 million reads and 8 million of those reads map to the target regions, the on-target ratio is calculated to be 80%. The remaining 20% of reads may represent off-target reads that align to other regions of the genome.
The on-target ratio is influenced by the size of the panel, the nature of the nucleotide sequence of the target region, and also the method and process for designing the probes. In general, the larger the size of target region, the better the panel’s on-target performance. On-target performance decreases when hard-to-capture regions such as extremely high or low GC contents, repeat elements, and highly homologous regions are included. For these areas, meticulous panel design and assay optimization is often required to ensure capture performance and sequencing efficiency.
3. Uniformity
Uniformity of targeted sequencing refers to the evenness of coverage across the targeted regions of interest in a sample. In other words, it measures how well the sequencing reads are distributed across the targeted regions, with the goal to achieve consistent coverage for all targeted regions.
High uniformity means that all targeted regions have a similar depth of coverage, which is desirable for accurate and reliable variant calling. Conversely, low uniformity indicates that some regions have much higher or lower coverage than others, which can lead to uneven variant detection and poor sensitivity. Uniformity is typically measured by the standard deviation value of the depth for each target regions, or the coverage value corresponding to 0.2X or 0.5X of the average depth.
Uniformity can be affected by several factors, including the target capture efficiency, PCR amplification bias, and sequencing instrument parameters. It is important to optimize these factors to achieve high uniformity in targeted sequencing experiments.
4. Reproducibility
This refers to the consistency of results across multiple runs of the panel. The reproducibility can be evaluated by measuring the coefficient of variation of coverage or variant calling metrics.
Overall, a comprehensive evaluation of the performance of an NGS target capture panel should be addressed using a combination of these metrics, as well as the application-specific nature and requirements of each panel.