R/PCA.R

In Kaleido-Biosciences/phgrofit: Extracting Physiological Data From Kinetic OD600 and pH Curves

#' PCA: Easily making a PCA plot with phgrofit data
#' Copyright (c) 2019. Kaleido Biosciences. All Rights Reserved
#'
#' Intended to allow the user to easily generate PCA plots from phgrofit data.
#' Note that the values should first be scaled before using this function.
#' This can be acomplished with scale_phgrofit()
#' @param phgrofit_data This is the output origniating from phgrofit or one
#' of it's modifying functions such as scale_phgrofit or avg_phgrofit.
#' @param group This is the name of the column that you would like to color the ordination by.
#' A 95 percent confidence interval for this group will be assigned.
#' @param mouse_over Category for which you would like data on mouse over on the plot.
#'
#' @return a ggplot2 object containing the 95 percent confidence interval for the group specified.
#' @export
#' @importFrom magrittr %>%
#' @examples
#' ### phgropro processing
#' \dontrun{phgropro_output = phgropro(biotek_export = filepath.txt,metadata = metadata.csv,Plate_Type = 96)}
#'
#' ### phgrofit processing
#' \dontrun{phgrofit_output = phgrofit(phgropro_output)}
#'
#' ## scaling phgrofit data
#'\dontrun{ phgrofit_data = scale_phgrofit(phgrofit_output)}
#'
#' ### printing PCA plot with colored confidence intervals for community
#' \dontrun{community_PCA = PCA(phgrofit_data,"Community")
#' print(community_PCA)}
PCA = function(phgrofit_data,group="Sample.ID",mouse_over = "Compound"){
    if("min_pH" %in% names(phgrofit_data)){
    #if it is phgrofit data
    #Selecting numeric data
    PCA_data = dplyr::select(phgrofit_data,
                             od600_lag_length,
                             od600_max_gr,
                             max_od600,
                             difference_between_max_and_end_od600,
                             auc_od600,
                             max_acidification_rate,
                             min_pH,
                             time_of_min_pH,
                             max_basification_rate,
                             max_pH,
                             difference_between_end_and_min_pH,
                             auc_pH
    )

    #PCA
    PCA_vals = prcomp(PCA_data,center = FALSE, scale = FALSE)

    #Selecting things that may be groups
    groups = dplyr::select(phgrofit_data,-c(od600_lag_length,
                                            od600_max_gr,
                                            max_od600,
                                            difference_between_max_and_end_od600,
                                            auc_od600,
                                            max_acidification_rate,
                                            min_pH,
                                            time_of_min_pH,
                                            max_basification_rate,
                                            max_pH,
                                            difference_between_end_and_min_pH,
                                            auc_pH)
    )
    }else{
    # else it is grofit data
    PCA_data = dplyr::select(phgrofit_data,
                             od600_lag_length,
                             od600_max_gr,
                             max_od600,
                             difference_between_max_and_end_od600,
                             auc_od600)
    #PCA
    PCA_vals = prcomp(PCA_data,center = FALSE, scale = FALSE)

    #Selecting things that may be groups
    groups = dplyr::select(phgrofit_data,-c(od600_lag_length,
                                            od600_max_gr,
                                            max_od600,
                                            difference_between_max_and_end_od600,
                                            auc_od600))

}
    #Selecting specified groups
    sel_group = dplyr::select(groups,group)

    #Getting PCA values
    PCA_df = predict(PCA_vals,PCA_data) %>%
        cbind(groups)

    #extracting proportion of varience explained by the PCAs
    Prop_of_var = data.frame(summary(PCA_vals)$importance)[2,]



    #Plotting with 95% confidence interval
    p1 = ggplot2::ggplot(PCA_df,ggplot2::aes_string("PC1","PC2",color = group))+
        ggplot2::geom_point(ggplot2::aes_string(text = mouse_over))+
        ggplot2::stat_ellipse(level=0.95)+
        ggplot2::theme(legend.position = "bottom")+
        ggplot2::ggtitle(paste0("PCA Plot Colored by ",group))+
        ggplot2::xlab(paste0("PCA-1 ",round(Prop_of_var[1]*100,2),"%"))+
        ggplot2::ylab(paste0("PCA-2 ",round(Prop_of_var[2]*100,2),"%"))+
        ggplot2::theme_bw()+
        ggplot2::theme(plot.title = ggplot2::element_text(hjust = 0.5))

    return(p1)
}