BayesCall: A model-based base-calling algorithm for high-throughput short-read sequencing

Click on image to view larger version.

Table 2.

Confusion matrices for the 76-cycle phiX174 data on GA-II

Click on table to view larger version.

Table 2.
  • The (x,y) entry in the table corresponds to the percentage of times that base x is called as base y. These tables illustrate that BayesCall is in general less “confused;” that is, the diagonal entries for BayesCall are higher than the corresponding entries for the other methods, except for the (T,T) entry in Alta-Cyclic. Further, BayesCall suffers less from the “anomalous T” effect than does Bustard or Alta-Cyclic; i.e., the (A,T), (C,T), and (G,T) entries for BayesCall are less than the corresponding entries for Bustard or Alta-Cyclic.

This Article

  1. Genome Res. 19: 1884-1895

Preprint Server