Noise2read: Accurately Rectify Millions of Erroneous Short Reads Through Graph Learning on Edit Distances
{{output}}
Although the per-base error rate of short-read sequencing data is very low at 0.1%-0.5%, the percentage/probability of erroneous reads in a dataset can be as high as 10%-15% or in the number millions. As current methods correct only some errors while introduci... ...
