1 Trimming Reads

Observing the nonuniform distribution of nucleotide frequencies surrounding the 5’end of reads, a natural step to take would be to trim the 5’end before mapping, in the hope that simply removing the portion of the read in which the bias occurs will also remove the bias. Figure 1 demonstrates that this is not the case. Trimming the initial heptamer in the Mortazavi data set does nothing to reduce the bias, and simply shifts the plot by seven positions. This indicates that the issue is sampling bias, rather than a bias in base calling.