Supporting Information and Figures.

Gene and primer sequences for tuberculosis and MRSA strains. Additional details for PCR and restriction digest (Fig A). Bayesian classification for Mode I: Additional data and numerical simulations (Figs B-E). Bayesian classification for Mode II: Additional data and numerical simulations (Fig F). M. tuberculosis and methicillin-resistant S. aureus SNV detection: Additional data and numerical simulations (Figs G and H). Fig A in S1 File. PCR and Restriction Digest Products. Native PAGE showing successful PCR amplification of the mazG gene (M. tuberculosis) and parC gene (S. aureus). Digestion reactions yield either cut or uncut amplicons depending upon the parent strain. Lane 1: loading dye. Lane 2: 100 bp NEB ladder. Lane 3: mazG gene amplified from H37Ra (942 bp). Lane 4: mazG gene amplified from H37Rv (942 bp). Lane 5: mazG from H37Ra after digestion with NaeI enzyme, not cut. Lane 6: mazG from H37Rv after digestion with NaeI, cut into two fragments of 321 and 621 bp. Lane 7: parC gene fragment amplified from HOU-MR strain (885 bp). Lane 8: parC gene fragment amplified from FPR3757 strain (885 bp). Lane 9: parC from HOU-MR after digestion reaction with BseRI, cut into two fragments of 245 and 640 bp. Lane 10: parC from FPR3757 after digestion reaction with BseRI, not cut. Digestion reactions were performed at 37°C for 1hr in NEB Cutsmart buffer. 10 units of enzyme were used for each digestion reaction. Fig B in S1 File. Gaussian Mixture Model Fits for DNA Translocation. Gaussian mixture model fits to translocations of single-length DNA samples through a 4.8 nm diameter nanopore (1M KCl, +300 mV bias). (a) 100 bp NoLimits DNA. (b) 200 bp NoLimits DNA. (c) 900 bp NoLimits DNA. (d) 1000 bp NoLimits DNA. Raw tD and ΔI data are shown in Fig 2 (main text). Fig C in S1 File. Bayesian Posterior Estimates for Nanopore Sample Identification. Bayesian posterior estimates p(100bp|Θ) and p(1000bp|Θ) for test data sets of N points given a model based on M points. Data is bootstrapped from translocations of (a) 100 bp NoLimits DNA and (b) 1000 bp NoLimits DNA (main text: Fig 2A and 2D) corresponding to the Gaussian Mixture Models shown in Figs Ba and Bd. Each point represents the average of 1000 simulated posterior estimates, each of which uses a randomly selected model set M and test set N. Fig D in S1 File. Mode I: Identification of 100 bp vs. 200 bp DNA. Bayesian posterior estimates p(100bp|Θ) and p(200bp|Θ) for test data sets of N points given a model based on M points. Data is bootstrapped from translocations of (a) 100 bp NoLimits DNA and (b) 200 bp NoLimits DNA (main text: Fig 2A and 2B) corresponding to the Gaussian mixture models shown in Figs Ba and Bb. Each point represents the average of 1000 simulated posterior estimates, each of which uses randomly selected (disjoint) model set M and test set N. Fig E in S1 File. Mode I: Identification of 900 bp vs. 1000 bp DNA. Bayesian posterior estimates p(900bp|Θ) and p(1000bp|Θ) for test data sets of N points given a model based on M points. Data is bootstrapped from translocations of (a) 900 bp NoLimits DNA and (b) 1000 bp NoLimits DNA (main text: Fig 2C and 2D) corresponding to the Gaussian mixture models shown in Figs Bc and Bd. Each point represents the average of 1000 simulated posterior estimates, each of which uses randomly selected (disjoint) model set M and test set N. Fig F in S1 File. Mode II: Identification of 1000 bp vs 800+200 bp DNA. (a) 1000 bp at 1 nM. (b) 1:1 ratio of 800 bp + 200 bp, total concentration 2 nM. (c) Gaussian mixture model fit, 1000 bp. (d) Gaussian mixture model fit, 800 bp + 200 bp. (e) Bayesian posterior estimate p(1000bp|Θ) for test data sets of N points given a model based on M points. (f) Bayesian posterior estimate p(800+200bp|Θ) for test data sets of N points given a model based on M points. Translocations for all samples were collected in a single nanopore (4.8 nm diameter, effective thickness ~7 nm) with a +300 mV bias relative to trans (open pore current: 13 nA). To facilitate visualization of population density, a random white noise offset below the acquisition rate of this data (-2 μs < Δt < +2 μs, acquisition rate 250 kHz) has been added to each tD in panels (a) and (b). Numerical simulations for panels (e) and (f) were bootstrapped from the data in panels (a) and (b), respectively. Each point represents the average of 1000 simulated posterior estimates, each of which uses randomly selected (disjoint) model set M and test set N. Fig G in S1 File. Identification of M. tuberculosis H37Ra vs. H37Rv mazG Samples. Bayesian posterior estimates p(H37Ra|Θ) and p(H37Rv|Θ) for test data sets of N points given a model based on M points. Data is bootstrapped from translocations of (a) Tuberculosis H37Ra and (b) H37Rv mazG restriction digested fragments as described in S1 File Sections 1 and 2. Each point represents the average of 1000 simulated posterior estimates, each of which uses randomly selected (disjoint) model set M and test set N. Fig H in S1 File. Identification of S. aureus FPR3757 vs. HOU-MR parC Samples. Bayesian posterior estimates p(FPR3757|Θ) and p(HOU-MR|Θ) for test data sets of N points given a model based on M points. Data is bootstrapped from translocations of (a) MRSA FPR3757 and (b) HOU-MR parC restriction digested fragments as described in S1 File Sections 1 and 2. Each point represents the average of 1000 simulated posterior estimates, each of which uses randomly selected (disjoint) model set M and test set N.
(PDF)