Enhancing Epitranscriptome Module Detection from m⁶A-Seq Data Using Threshold-Based Measurement Weighting Strategy

Figure 3

Parameter optimization for the threshold-based method. Small datasets are generated by sampling randomly from real RNA methylation dataset, to which clustering analysis used the threshold-based weighting strategy with different parameters and the clustering performance was evaluated by comparing to true sample labels. Better clustering performance was achieved when setting a relative small value for weight parameter and a medium value for threshold parameter .

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)