Understanding sequence errors in analysis of short tandem repeats through optimisation of a library preparation protocol utilising unique molecular indices

Abstract: In the last few years massively parallel sequencing (MPS) has become important for forensic DNA analysis by providing new and broader possibilities, such as increased discrimination power for short tandem repeat (STR) analyses. An imperative part of MPS analysis is library preparation where there is a fundamental reliability on in vitro DNA polymerisation, which is prone to errors. In this project, a two-step polymerase chain reaction-based (PCR) library preparation protocol which utilises unique molecular indices was optimised for analysis of STRs. By investigating the effects of varying DNA polymerases with different properties, annealing temperature and cycles, the study aimed to improve the understanding of the nature of polymerisation errors and the effect on STR-MPS sequencing data. It was observed that the type of DNA polymerase used in the barcode-incorporating PCR step greatly impacts the generation of target specific reads. Platinum SuperFi II generated an average of 51% reads recognised to an STR marker while Phusion Hot Start II HF generated an average of 8% recognised reads. Furthermore, we found that increasing the annealing temperature from 58°C to 62°C resulted in an average reduction of 57% of the target specific yield while the proportion of the reads that generated the correct genotype was observed at above 94% for all samples. The protocol was evaluated by analysing two-person DNA mixtures of different proportions. It was shown that the correct genotype of the minor contributor was detected down to 1:49 proportion mixtures with 5ng total DNA template input. In conclusion, it has been shown that the type of DNA polymerase and PCR conditions greatly impact the target specific yield but that the incorporation of unique molecular indices (UMIs) effectively reduces sequence errors to ~5%, regardless of reaction conditions. These results have generated a better understanding of amplification efficiency and sequencing errors and may be applied to guide future optimisation of MPS assays.