The origin of biased sequence depth in sequence-independent nucleic acid amplification and optimization for efficient massive parallel sequencing

Sequence Independent Single Primer Amplification is one of the most widely used random amplification approaches in virology for sequencing template preparation. This technique relies on oligonucleotides consisting of a 39 random part used to prime complementary DNA synthesis and a 59 defined tag sequence for subsequent amplification. Recently, this amplification method was combined with next generation sequencing to obtain viral sequences. However, these studies showed a biased distribution of the resulting sequence reads over the analyzed genomes. The aim of this study was to elucidate the mechanisms that lead to biased sequence depth when using random amplification. Avian paramyxovirus type 8 was used as a model RNA virus to investigate these mechanisms. We showed, based on in silico analysis of the sequence depth in relation to GC-content, predicted RNA secondary structure and sequence complementarity to the 39 part of the tag sequence, that the tag sequence has the main contribution to the observed bias in sequence depth. We confirmed this finding experimentally using both fragmented and non-fragmented viral RNAs as well as primers differing in random oligomer length (6 or 12 nucleotides) and in the sequence of the amplification tag. The observed oligonucleotide annealing bias can be reduced by extending the random oligomer sequence and by in silico combining sequence data from SISPA experiments using different 59 defined tag sequences. These findings contribute to the optimization of random nucleic acid amplification protocols that are currently required for downstream applications such as viral metagenomics and microarray analysis