Massively parallel sequencing of micro-manipulated cells targeting a comprehensive panel of disease-causing genes: A comparative evaluation of upstream whole-genome amplification methods

Abstract

Single Gene Disorders (SGD) are still routinely diagnosed using PCR-based assays that need to be developed and validated for each individual disease-specific gene fragment. The TruSight One sequencing panel currently covers 12 Mb of genomic content, including 4813 genes associated with a clinical phenotype. When only a limited number of cells are available, whole genome amplification (WGA) is required prior to DNA target capture techniques such as the TruSight One panel. In this study, we compared 4 different WGA methods in combination with the TruSight One sequencing panel to perform single nucleotide polymorphism (SNP) genotyping starting from 3 micro-manipulated cells. This setting simulates clinical settings such as day-5 blastocyst biopsy for Preimplantation Genetic Testing (PGT), liquid biopsy of circulating tumor cells (CTCs) and cancer-cell profiling. Bulk cell samples were processed alongside these WGA samples to serve as a performance reference. Target coverage, coverage uniformity and SNP calling accuracy obtained using any of the WGA, is inferior to the results obtained on bulk cell samples. However, results after REPLI-g come close. Compared to the other WGA methods, the method using REPLI-g WGA results in a better coverage of the targeted genomic regions with a more uniform read depth. Consequently, this method also results in a more accurate SNP calling and could be considered for clinical genotyping of a limited number of cells.

Fig 1. Experimental design.

Three cell samples from the NA12882 cell line were amplified with either Ampli1 (3 replicates), REPLI-g (3 replicates), SurePlex (5 replicates) or MALBAC (3 replicates). Subsequently, samples were randomized for TruSight One capture in pools of 3 and those pools were sequenced on separate MiSeq runs. Four bulk DNA sample from the NA12882 cell line, not amplified before capture, were also randomly included in the pools.

Fig 2. Read distribution.

(A) Read depth calculated in 1 kb sliding windows across the concatenated target regions for one sample of each method. (B, C) Lorenz curves and Gini indexes describing how sequenced bases are distributed over targeted (B) and covered (C) bases.

Fig 3. Percent target coverage at various minimum read depths.