1/13 While #COVID19 warrants our attention and swift, concerted action, we cannot forget about the existing high global burden diseases like #tuberculosis.
2/13 In ”Exact mapping of Illumina blind spots in the Mycobacterium tuberculosis genome reveals platform-wide and workflow-specific biases” preview.tinyurl.com/vyhckyc, @crobinhold1 @Valafarlab & I provide a roadmap for navigating #Illumina coverage bias when #sequencing Mtb #genomes
3/13 We use a #phylogenomic filtering step to remove evolutionary signals isolating systematically low coverage and provide lists of #Illumina #blindspots for seven #sequencing workflows (lib prep + instrument) at single-base resolution
4/13 A big takeaway is that many positions of PE/PPE genes are not blind spots and can therefore be confidently sequenced. Many researchers preclude them from short-read Mtb sequencing because they are repetitive/GC-rich
5/13 Conversely, ~45% of blind spots were in non PE/PPE genes, affecting 529 genes! This included some genes associated with drug #Resistance. #AMR
6/13 Across the seven workflows there was a clear standout: a modified #Nextera library prep that employs a high-fidelity polymerase during PCR, reducing amplification bias
7/13 Looking at how bias varied across sequence composition made the source of reduced bias clear: blind spots in GC-rich (prone to PCR bias) regions were markedly less common w/mod Nextera lib prep, while repeats were similar (which are lower coverage due to mapping ambiguity)
8/13 When investigating which features of sequence composition gave #Illumina the most trouble, we found a surprise: #Homopolymers shorter than expected (6 bp, perhaps lower) caused systematic coverage bias, shorter than commonly reported.
9/13 The bias in #homopolymers persisted after excluding homopolymeric positions that coincided with seqeunce features known to drive bias (GC-rich or tandem repeats)
10/13 The modified Nextera lib prep with high-fidelity PCR polymerase also performed markedly better on these isolated homopolymeric positions, suggesting polymerase slippage during amplification drove some of the homopolymer bias
11/13 Lastly, some of the blind spots were salvageable, while others weren’t: position-specific coverage increased with increased genome-wide coverage (see Fig. 6 for examples).
12/13 We think this can serve as a resource for the #tuberculosis research community to use when designing #WGS experiments and for calibrating position-specific skepticism for studies using published #Illumina WGS data, e.g. large-scale #GWA studies
13/13 And to be clear, this is not a knock on #Illumina sequencing, but rather an exercise to empirically evaluate how to handle the growing abundance of M. tuberculosis Illumina WGS data
biorxiv.org/content/10.110… (Direct link)
