Pc Miler 26 Torrent Hit 
Pc Miler 26 Torrent Hit
a mock community sample was prepared from dna extracts of five strains of four bacterial species for sequencing comparison, and a set of ion torrent adapters (table s7 and table s8) was designed accordingly. these were sequenced on the ion torrent pgm.
different samples were then processed on different platforms, dna was extracted, adapters were ligated, and bar codes were sequenced on different platforms, including the ion torrent pgm and the ion torrent proton. initial read identification, quality assessment, and error rate calculations were done with seqtrace [ 5 ] (for pgm) and torrent suite v3.1.1 (for proton). for the bacterial sample, we applied micra for all treatment types. for the mock community sample, micra was run only twice: one time with the treatment based on the pgm protocol, and once with the treatment based on the ion torrent protocol. in addition, two artificial samples were processed with micra using the ion torrent protocol. the treatment and sequencing methods are summarized in table s9. in the protocol, we included a detailed description of the pre- and post-processing steps.
interestingly, in the case of the is1 transposase, the product of this transposase has previously been associated with susceptibility to colistin, a known mcr-1 variant [ 24 ]. in our case, the snv was found in a transposase (locus: ecdh10b_1059), among other is1 transposases, at very high frequency (90.14%) and with a depth of 96 reads. this suggests that this transposase or the neighboring is1 transposase might be associated with the is1 transposase variant causing the transfer of mcr-1 across different species  and putative mcr-1 plasmid .
sorting the sequences and counting them per bar code, a total of 2,582,538 reads were obtained, corresponding to 2,882,582 unique sequences for the 96 bar code library (204,051 reads) and 2,091,175 reads for the 384 bar code library (311,023 reads). reads were mapped to the b73 version of the b73 refgen_v2 reference genome (table s3). for each bar code, different sets of samples were used resulting in different read number and sequencing depth for each sample. samples were sorted by bar code and trimmed to remove bar code, sequencing adapter, and low-quality sequence. the mapping information was used to create lookup tables for read number to bar code. for sets of 96 and 384 bar codes, the read number per bar code ranged from 266 to 72,912 (96 bar codes) and 6,347 to 437,371 (384 bar codes), respectively. on average, 20% of the reads were mapped to the genome. after quality trimming, the average read length of the best mapping location was 126 bp (96 barcodes) and 116 bp (384 barcodes). to determine the percentage of fragments that had bar codes added, each sample was split into two sets, those that had a bar code and those that didn’t. for each bar code, each set was mapped to the genome and the number of bar coded fragments was compared to the total number of fragments. in the 96 bar code library, the ratio of bar coded fragments ranged from 85 to 96%. in the 384 bar code library, the ratio was 92.5%. an exact binomial test was applied to determine the probability of seeing specific read counts for each bar code. counts of reads at all read depth were calculated, and the corresponding probability was determined. 5ec8ef588b