Discovering candidates for gene network expansion by variable subsetting and ranking aggregation