Bias reduction improves accuracy and informativity of high-throughput sequencing chimerism assays

Chimerism monitoring by means of high-throughput sequencing of biallelic polymorphisms has shown promising advantages for patient follow-up after hematopoietic stem cell transplantation. Yet, the presence of method bias precludes achievement of an assay's theoretically attainable informativity rate, as method bias necessitates the exclusion of some markers. This method bias arises because of preferential observation of one allele over the other, and for some allelic constellations because of stochasticity.

This paper suggests how preferential allelic observation may lead to method bias, and when and why such bias necessitates the exclusion of markers. It is shown that also markers that remain informative suffer a reduction in trueness and precision due to method bias. A bias reduction approach in the data analysis phase is introduced and shown to improve trueness and precision under all circumstances, meriting its universal adoption. This bias reduction furthermore allows to achieve an assay's theoretically achievable informativity rate, though at the cost of reduced sensitivity. Several strategies to consider in the assay design phase that may lower biases are proposed.

Improved design and data analysis of chimerism assays increase the accuracy, applicability, and cost-effectiveness of high-throughput sequencing chimerism assays.