Optimising scalability
On average 30 targets per MinION run were amplified, pooled and sequenced. Initially, a limited number of 11 samples were barcoded and sequenced, but a reduction in usable data was noted (Supplementary Table 3). As such, this approach was replaced with careful sample pooling instead. The 11 samples were not re-sequenced, as the small number meant there was still ample data per target (mean of 19 million bases per target). Targets were checked to ensure no overlap in genome location, and demultiplexing was performed based on region extraction of the mapped BAM files (see methods ‘Bioinformatics’). Using this approach, a loss of 17.8% (SEM 7.2) of total basecalled yield was avoided by bypassing barcode ligations, de-multiplexing binning issues and barcode trimming (Supplementary Table 3).