Fig. 3 | Nature Communications

Fig. 3

From: Community assessment to advance computational prediction of cancer drug combinations in a pharmacogenomic screen

Fig. 3

Performance of teams in the AstraZeneca-DREAM challenge. a Participant performance in SC1A and SC1B— the distribution of performance of random predictions was used to estimate a lower limit, and the distribution of synergy correlations between biological replicates were used to estimate the upper limit. b Participant performance ranked in SC2 based on the primary metric, 3-way ANOVA. Distribution of bootstrap prediction performances for each team are shown by each boxplot with the dot showing their actual performance. c Participant performance plotted with upper and lower limits for SC2 based on the tie-break metric. d Ensemble models compared to the performance of individual models ranked from best to poorest performing in SC2. SML is an ensemble of the best performing models based on estimation of their BAC. Random Aggregation is an ensemble combining a random combination of models. Standard error of mean represented by error bars are estimated from ten random splits of the data

Back to article page