Here, we pursued a framework-LncMod,
for identifying the lncRNA modulator by integrating genome-wide
gene expression profiles and transcription regulations. This
process invovled several scoring and filtering steps, as
illustrated in Fig. 1. The proposed method takes four inputs:
the gene expression profile dataset for lncRNA, TF and target
genes, the context-specific TF-gene regulations. For each
TF-gene regulations, we reported the lncRNA modulators along
with their mode of action.
I) Briefly, the paired lncRNA and gene expression
profiles of a specific cancer were obtained and the lncRNA, TF
and genes were filtered based on the expression variation
across samples ('range constraint').
II) In addition, the expressions of the candidate lncRNA
modulator and of the TF are required to be statistically
independent ('independence constraint'). And then the estimator
assesses the statistical significance of the difference in
correlation between the TF and a target in two subsets-the top
and bottom 25% of samples in which the candidate lncRNA
modulator is most and least expressed.
III) For each possible lncRNA-TF-gene triplet is
independently tested using the permutation method. False
positives are controlled using appropriate statistical
thresholds. |
