aggregateRanks {miRNAtap} | R Documentation |
This function performs aggregation phase of target
prediction for getPredictedTargets
.
Consensus ranking is derived from multiple individual rankings.
Available methods include minimum, maximum and geometric mean with further
tuning parameters which promote true positives at the top of the final
ranking
aggregateRanks(ranks, n_valid_srcs, min_src, method = "geom", promote = TRUE)
ranks |
|
n_valid_srcs |
number of valid sources in the dataset |
min_src |
minimum acceptable number fo sources |
method |
|
promote |
add weights to improve accuracy of the method, default
|
data.frame
object with ranks per source and aggregate ranks
Maciej Pajak m.pajak@sms.ed.ac.uk
data = data.frame(GeneID=c("15364", "56520", "57781", "58180", "18035"), source1scores=c(0.9,0.5,0.3,NA,NA), source2scores=c(0.7,NA,0.8,0.6,0.5), source3scores=c(0.5,NA,0.3,0.1,0.2)) data #dataframe with scores aggregateRanks(data, n_valid_srcs=3, min_src=2, method='geom') #note how gene 56520 is eliminated as it appeared in fewer than 2 sources