Limits...
A crowd-sourcing approach for the construction of species-specific cell signaling networks.

Bilal E, Sakellaropoulos T, Melas IN, Messinis DE, Belcastro V, Rhrissorrakrai K, Meyer P, Norel R, Iskandar A, Blaese E, Rice JJ, Peitsch MC, Hoeng J, Stolovitzky G, Alexopoulos LG, Poussin C, Challenge Participan - Bioinformatics (2014)

Bottom Line: Such a large network inference challenge not based on synthetic simulations but on real data presented unique difficulties in scoring and interpreting the results.Because any prior knowledge about the networks was already provided to the participants for reference, novel ways for scoring and aggregating the results were developed.Supplementary data are available at Bioinformatics online.

View Article: PubMed Central - PubMed

Affiliation: IBM Research, Computational Biology Center, Yorktown Heights, NY 10598, USA, ProtATonce Ltd, Scientific Park Lefkippos, Patriarchou Grigoriou & Neapoleos 15343 Ag. Paraskevi, Attiki, Greece, National Technical University of Athens, Heroon Polytechniou 9, Zografou, 15780, Greece and Philip Morris International R&D, Philip Morris Products S.A., Quai Jeanrenaud 5, 2000 Neuchâtel, Switzerland.

Show MeSH
(A) The beta-binomial mixture weight can be calculated by maximizing the log-likelihood function. (B) Using this value, the fitted mixture is shown in red together with the individual-weighted components in black. Only edges present in the reference network were used in this case
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4325542&req=5

btu659-F4: (A) The beta-binomial mixture weight can be calculated by maximizing the log-likelihood function. (B) Using this value, the fitted mixture is shown in red together with the individual-weighted components in black. Only edges present in the reference network were used in this case

Mentions: The optimal threshold for building the consensus network was determined by fitting the model described in Section 2 (Equations 3, 4, and 5) followed by solving Equation 6. The data used for the fit were assembled by counting the number of ‘votes’ received by each edge in the reference network from the participating teams (excluding Team 93) and the silver standard network. This was performed separately for human and rat networks, and then the resulting datasets were mixed to improve the fit. Maximizing the log likelihood function of the mixture of two beta-binomial distributions (Equation 3) for different mixing constants led to Pr(Y = 1) = 0.16, Pr(Y = 0) = 0.84 (Fig. 4A) and shape parameters a1 = 8.77e+06, b1 = 1.95e+06, a2 = 3.46e+06 and b2 = 1.57e+06. Using this result and after solving Equation 6, it was found that it takes approximately eight votes to verify the condition Pr(Y = 1/X = k) > Pr(Y = 0/X = k). This result can also be visualized in Figure 4B by tracing the intersection of the two mixture components depicted in black.Fig. 4.


A crowd-sourcing approach for the construction of species-specific cell signaling networks.

Bilal E, Sakellaropoulos T, Melas IN, Messinis DE, Belcastro V, Rhrissorrakrai K, Meyer P, Norel R, Iskandar A, Blaese E, Rice JJ, Peitsch MC, Hoeng J, Stolovitzky G, Alexopoulos LG, Poussin C, Challenge Participan - Bioinformatics (2014)

(A) The beta-binomial mixture weight can be calculated by maximizing the log-likelihood function. (B) Using this value, the fitted mixture is shown in red together with the individual-weighted components in black. Only edges present in the reference network were used in this case
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4325542&req=5

btu659-F4: (A) The beta-binomial mixture weight can be calculated by maximizing the log-likelihood function. (B) Using this value, the fitted mixture is shown in red together with the individual-weighted components in black. Only edges present in the reference network were used in this case
Mentions: The optimal threshold for building the consensus network was determined by fitting the model described in Section 2 (Equations 3, 4, and 5) followed by solving Equation 6. The data used for the fit were assembled by counting the number of ‘votes’ received by each edge in the reference network from the participating teams (excluding Team 93) and the silver standard network. This was performed separately for human and rat networks, and then the resulting datasets were mixed to improve the fit. Maximizing the log likelihood function of the mixture of two beta-binomial distributions (Equation 3) for different mixing constants led to Pr(Y = 1) = 0.16, Pr(Y = 0) = 0.84 (Fig. 4A) and shape parameters a1 = 8.77e+06, b1 = 1.95e+06, a2 = 3.46e+06 and b2 = 1.57e+06. Using this result and after solving Equation 6, it was found that it takes approximately eight votes to verify the condition Pr(Y = 1/X = k) > Pr(Y = 0/X = k). This result can also be visualized in Figure 4B by tracing the intersection of the two mixture components depicted in black.Fig. 4.

Bottom Line: Such a large network inference challenge not based on synthetic simulations but on real data presented unique difficulties in scoring and interpreting the results.Because any prior knowledge about the networks was already provided to the participants for reference, novel ways for scoring and aggregating the results were developed.Supplementary data are available at Bioinformatics online.

View Article: PubMed Central - PubMed

Affiliation: IBM Research, Computational Biology Center, Yorktown Heights, NY 10598, USA, ProtATonce Ltd, Scientific Park Lefkippos, Patriarchou Grigoriou & Neapoleos 15343 Ag. Paraskevi, Attiki, Greece, National Technical University of Athens, Heroon Polytechniou 9, Zografou, 15780, Greece and Philip Morris International R&D, Philip Morris Products S.A., Quai Jeanrenaud 5, 2000 Neuchâtel, Switzerland.

Show MeSH