Skip to main content

Advertisement

Table 2 Comparison of the performances by different machine learning algorithms

From: Improving ChIP-seq peak-calling for functional co-regulator binding by integrating multiple sources of biological information

  Number of peaks Number of peaks with ERE motif Ratio of peaks with ERE motif match
MACS p = 1E-10 1,286 941 0.73
MACS p = 1E-8 1,966 1,416 0.72
MACS p = 1E-5 4,678 3,077 0.66
k-means (city block)
Cluster 1 26,211 11,943 0.46
Cluster 2 12,113 3,245 0.27
supervised-NB(th = 0.8,1:2)
Positively labeled 11,835 8,196 0.69
Negatively labeled 26,489 6,992 0.26
supervised-SVM(kernel = polynomal,1:2)
Positively labeled 14,915 8,425 0.56
Negatively labeled 23,409 6,763 0.29
supervised-RF(th = 0.7,1:2)
Positively labeled 10,428 6,514 0.62
Negatively labeled 27,896 8,674 0.31
semi-supervised-NB(th = 0.8,1:2, I = 75)
Positively labeled 12,597 8,458 0.67
Negatively labeled 25,727 6,730 0.26