Significance of the contractions and expansions along branches was accessed by means of the three methodologies available in CAFE: branch cutting, likelihood ratio test, and Viterbi assignments [122].
There was no statistically significant difference between the average values obtained by the two methodologies for either refined oils or olive oils.
However, the final list of genes comprised 86 probe sets (74 genes) commonly yielded by the two methodologies which were differentially expressed in LGMD2A compared to unaffected muscle biopsies.
Thus, while some RNA-seq and ChiP-seq targets were in common, there was not strict concordance between targets identified by the two methodologies.
We directly compared the CoNECT dataset (Table S16) with the RNA-seq dataset by Gan et al. (2010) genome-wide and generated a scatter plot of FPKM values produced by the two methodologies.
AMICA1 (#48), MMP9 (#2), TYROBP (#49), SELPG (# 604), LEF1 (#64) were selected as candidate genes to compare the fold over-expression by the two methodologies (# represents fold over-expression relative to B cell based on FPKM data, smaller number indicates higher fold overexpression).
Table 7 Performances obtained using Euclidean distance by combining the two methodologies with the two sets of reference (threshold with α=2) ComplOverlap Eucl + All Output×G T 1 70 83 Output×G T 2 77 83.
Table 1 Performances obtained with Canberra distance by combining the two methodologies with the two reference sets (threshold with α=1) ComplOverlap Canb + All Output×G T 1 50 57 Output×G T 2 43 47.
Table 4 Performance obtained with statistical value χ 2 by combining the two methodologies with two sets of reference (threshold with α=1) ComplOverlap χ 2 + All Output×G T 1 50 47 Output×G T 2 43 43.
Table 6 Performances obtained using Euclidean distance by combining the two methodologies with the two sets of reference (threshold with α=1) ComplOverlap Eucl + All Output×G T 1 50 43 Output×G T 2 43 47. Figure 14 shows the results from using thresholds with α=2.
