Limitations

Data

LightSpeed is developed for and has been optimized on Illumina paired-end short read sequencing data. Paired-end sequencing data from other platforms utilizing the same data structure and similar read lengths can be expected to perform equally well with LightSpeed unless the background error-rate is markedly different. Analysis of other types of sequencing reads may not result in similar processing times or variant calls of an equivalent quality. Reads that are longer than 800 base pairs cannot be processed.

Variant detection

The germline variant detection algorithm in LightSpeed is based on a model expecting diploid genomes. Therefore, LightSpeed cannot be expected to accurately detect germline variants in genomes with other ploidies. In addition, alternate ploidies of sex chromosomes are not considered in the variant detection algorithm.

Somatic variant detection with LightSpeed is possible for variants down to a variant allele frequency of 0.1%. Variants below this frequency will not be considered. However, in order to ensure high accuracy in variant calling, we recommend only calling variants down to a variant allele frequency of approx. 1%.

Reference sequence

LightSpeed considers all chromosomes to be linear. Hence, for read mapping, circular chromosomes are linearized with position 1 starting at the junction of the chromosome. No reads will be mapped accross the junction of circular chromosomes.

UMI grouping

Duplex UMI handling is not supported, as reads must originate from the same strand to be grouped by LightSpeed.