selfea (version 1.0.1)

top_table: top_table

Description

Get IDs that pass two filters, p-value and effect-size. This top_table will make a significant list that is less than p-value and greater than effect-size. Effect-size are calculated by obtained power level. This function requires four parameters. ex) top_table(input_data,pvalue=0.05,power_desired=0.90,method='QPF')

Usage

top_table(input_list, pvalue = 0.05, power_desired = 0.9, method = "QPF",
  FC_threshold = 2)

Arguments

input_list

The list should be produced by 'get_statistics_from_file' or 'get_statistics_from_dataFrame' function. See get_statistics_from_file and

pvalue

p-value should be ranged between 0 to 1. default is 0.05.

power_desired

Give the statistical power you desired for output significant list

method

Choose statistics method you want to use for making significant list ll{ "QPF" combination of Quasi-Poisson and Cohen's f. Default. "QPF2" combination of Quasi-Poisson and Cohen's f2. "QPFC" combination of Quasi-Poisson and Fold ch

FC_threshold

Fold change you want to use. Default is 2.

Value

• A list containing the follow items and a scatter plot that x-axis is effect size and y-axis is probability. Vertical line the plot is minimum effect size and horizontal line is maximum probability threshold. Red dots means insignificant, while blue dots are significant. ll{ top_table a data frame that have calculated statistics for top table IDs minimum_effect_size Minimum effect size threshold selected_effect_size_filter The selected effect size filter minimum_power Minimum statistical power in the top_table selected_model The selected probability model for calculating p-value alpha Maximum adjusted p-value method_pvalue_adjustment The selected method for p-value adjustment num_group The number of groups used for generating the top_table common_replicates The number of common replicates. num_columns The number of columns (samples or experiments) } ll{ top_table's elements Cohens_W Cohen's w Cohens_F Cohen's f Cohens_F2 Cohen's f2 Max_FC Maximum fold change among all the possible group pairs QP_Pval_adjusted Adjusted p-value from GLM quasi-Poisson NB_Pval_adjusted Adjusted p-value from GLM negative binomial Normal_Pval_adjusted Adjusted p-value from Normal ANOVA }

Examples

library(selfea)

## Test selfea for single protein expression
values <- c(6,8,10,29,26,22)
groups <- c("U200","U200","U200","U600","U600","U600")
experiments <- c("exp1","exp2","exp3","exp4","exp5","exp6")

df_expr <- data.frame(ID="Protein_1",exp1=6,exp2=8,exp3=10,exp4=29,exp5=26,exp6=22)
df_group <- data.frame(Col_Name=experiments,Group=groups)
list_result <- get_statistics_from_dataFrame(df_expr,df_group)
top_table(list_result)

## For this example we will import Gregori data
## Josep Gregori, Laura Villareal, Alex Sanchez, Jose Baselga, Josep Villanueva (2013).
## An Effect Size Filter Improves the Reproducibility
## in Spectral Counting-based Comparative Proteomics.
## Journal of Proteomics, DOI http://dx.doi.org/10.1016/j.jprot.2013.05.030')

## Description:
## Each sample consists in 500ng of standard yeast lisate spiked with
## 100, 200, 400 and 600fm of a mix of 48 equimolar human proteins (UPS1, Sigma-Aldrich).
## The dataset contains a different number of technical replimessagees of each sample

## import Gregori data
data(example_data1)
df_contrast <- example_data
df_group <- example_group

## Get statistics through 'get_statistics_from_dataFrame' function
list_result <- get_statistics_from_dataFrame(df_contrast,df_group)

## Get significant features (alpha >= 0.05 and power >= 0.90)
significant_qpf <- top_table(list_result,pvalue=0.05,power_desired=0.90,method='QPF')