swing: Swing statistic
In KinSwingR: KinSwingR: network-based kinase activity prediction

Description Usage Arguments Value Examples

View source: R/swing.r

This function integrates the kinase-substrate predictions, directionality of phosphopeptide fold change and signficance to assess local connectivity (swing) of kinase-substrate networks. The final score is a normalised and weighted score of predicted kinase activity. If permutations are selected, network node:edges are permutated. P-values will be calculated for both ends of the distribution of swing scores (positive and negative swing scores).

1
2
3

swing(input_data = NULL, pwm_in = NULL, pwm_scores = NULL,
  pseudo_count = 1, p_cut_pwm = 0.05, p_cut_fc = 0.05,
  permutations = 1000, return_network = FALSE, verbose = FALSE)

`input_data`	A data.frame of phoshopeptide data. Must contain 4 columns and the following format must be adhered to. Column 1 - Annotation, Column 2 - centered peptide sequence, Column 3 - Fold Change [-ve to +ve], Column 4 - p-value [0-1]. This must be the same dataframe used in scoreSequences()
`pwm_in`	List of PWMs created using buildPWM()
`pwm_scores`	List of PWM-substrate scores created using scoreSequences()
`pseudo_count`	Pseudo-count acts at two levels. 1) It adds a small number to the counts to avoid zero divisions, which also 2) avoids log-zero transformations. Note that this means that pos, neg and all values in the output table include the addition of the pseudo-count. Default: "1"
`p_cut_pwm`	Significance level for determining a significant kinase-substrate enrichment. Default: "0.05"
`p_cut_fc`	Significance level for determining a significant level of Fold-change in the phosphoproteomics data. Default: "0.05"
`permutations`	Number of permutations to perform. This will shuffle the kinase-subtrate edges of the network n times. To not perform permutations and only generate the scores, set permutations=1 or permutations=FALSE. Default: "1000"
`return_network`	Option to return an interaction network for visualising in cystoscape. Default = FALSE
`verbose`	Turn verbosity on/off. To turn on, verbose=TRUE. Options are: "TRUE, FALSE". Default=FALSE

A data.table of swing scores

## import data
data(example_phosphoproteome)
data(phosphositeplus_human)

## clean up the annotations
## sample 100 data points for demonstration
sample_data <- head(example_phosphoproteome, 100)
annotated_data <- cleanAnnotation(input_data = sample_data)

## build the PWM models:
set.seed(1234)
sample_pwm <- phosphositeplus_human[sample(nrow(phosphositeplus_human), 
1000),]
pwms <- buildPWM(sample_pwm)

## score the PWM - substrate matches
## Using a "random" background, to calculate the p-value of the matches
## Using n = 100 for demonstration
## set.seed for reproducibility
set.seed(1234)
substrate_scores <- scoreSequences(input_data = annotated_data,
                                   pwm_in = pwms,
                                   background = "random",
                                   n = 100)

## Use substrate_scores and annotated_data data to predict kinase activity.
## This will permute the network node and edges 10 times for demonstration.
## set.seed for reproducibility
set.seed(1234)
swing_output <- swing(input_data = annotated_data,
                      pwm_in = pwms,
                      pwm_scores = substrate_scores,
                      permutations = 10)