Journal: Nature biotechnology
Article Title: De novo sequencing and variant calling with nanopores using PoreSeq
Figure Lengend Snippet: a) Accuracy results of running our code on nanopore data from M13, λ, and E. coli DNA to obtain complete de novo sequences. For M13, error bars indicate the upper and lower bounds for accuracy across 20 random subsets at the given coverage. The green line is the result of error correction and assembly with PBcR using only the 2D basecalled sequences; the red line shows the improvement when we error-correct with the raw data. b) Fraction of single-base variants of M13mp18 correctly called as a function of coverage. Variant sequences were generated by computationally making every possible insertion, deletion, or mutation in the original sequence of M13. A correct call is defined as the original M13 sequences' likelihood being larger than the variant in question. Error bars denote the deviation across 20 random subsets of molecules. c) Variant calling performance of our code on substitution mutations introduced in M13 at a higher frequency of 1%, at a range of coverages. Precision and recall denote the probabilities of false positives and negatives, respectively. The maximum F -score accuracy shown is 99.1% at 16× coverage.
Article Snippet: M13 Restriction Digest Four micrograms of M13mp18 RFI (New England Biolabs, cat. no. N4018S) DNA were digested with EcoRI restriction enzyme in a 100 microliter reaction volume for 2 hrs at 37C, and then heated for 30 min at 65C to inactivate the enzyme.
Techniques: Variant Assay, Generated, Mutagenesis, Sequencing