Table of Contents Author Guidelines Submit a Manuscript
BioMed Research International
Volume 2014, Article ID 351095, 12 pages
Research Article

Pathway Bridge Based Multiobjective Optimization Approach for Lurking Pathway Prediction

1Electrical and Computer Engineering Department, Texas A&M University, College Station, TX 77840, USA
2Radiology Comprehensive Cancer Center Cancer Biology, Wake Forest University, Winston-Salem, NC 27103, USA

Received 28 January 2014; Accepted 16 March 2014; Published 16 April 2014

Academic Editor: Xing-Ming Zhao

Copyright © 2014 Rengjing Zhang et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Supplementary Material

Supplementary 1.1: The figure showed the whole FANMOD analysis of the occurring frequency of all four-vetex motifs in the protein-to-protein network. Motifs ID 31710, 13278, 4958 and 27030 all had tiny frequency less than 5%, positive Z-score and zero p-value, which means they happened in smaller probabilities than the other structures in this network. Other than umbrella-looked ID 4382 and chain-looked ID 8598, the previous four motifs shared a common character—looping structure. Hence, the whole protein-to-protein network can be refined by triangle and rectangle structures, which are the only two possible looping elements of four-vetex motifs.

Supplementary 1.2: The table listed the top 30 predicted protein paths connecting miR205 and OCIAD2 ranked by the costs. The prediction cost was defined as the p-values summation of the proteins on the path. As p-value represents the probability that a random path has more occurring than the predicted one, the prediction with less cost was more reliable than others. However, due to the fact that the protein-to-protein interactions are undirected, calculation-based predictions might have paths with wrong directions that will never happen in the real biomedical world. Filtering followed this calculated result by deleting those bio-meaningless pathways.

Supplementary 1.3: In order to find the transcription factor of OCIAD2, the 177 human transcription factors were tried as the direct upstream of OCIAD2 one by one, and all the possible pathways were calculated and ranked by the prediction cost. The table showed the most possible 30 pathways. AR appeared in 19 out of the top 30 predictions; among them, AR worked as transcription factor of OCIAD2 for 9 times, followed by STAT5A with 3 times. Hence, among all the 177 human transcription factors, AR might be the most possible one for OCIAD2 in TGFβ signal.

  1. Supplementary Materials