In wilsontom/pickeR: Peak picking for low resolution liquid chromatography data

knitr::opts_chunk$set(echo = TRUE)
library(pickeR)
library(magrittr)
library(kableExtra)

Installation

To install using devtools;

devtools::install_github("wilsontom/pickeR")

or from source;

git clone https://github.com/wilsontom/pickeR
R CMD build pickeR
R CMD INSTALL pickeR_0.1.0.tar.gz

Introduction

Peak Smoothing

Peak smoothing performed using a Savitzky-Golay filter. The filter length is automatically determined for each chromatogram by calculting the smoothed intensity for a series of possible filter lengths. The selected filter length is determined by the minimum smoothed intensity being greater than the estimated noise of the chromatogram. This is to avoid the introdcution of too many negative intensity values.

rt <- chrom_181[[2]]$rt
int <- chrom_181[[2]]$int

smooth_int_opt <- smooth_peak(rt,int, ford = 4)

original <- plot_chrom(rt,int, peak_info = NULL)
original$labels$title <- 'Original Peak'

smoothed <- plot_chrom(rt,smooth_int_opt, peak_info = NULL)
smoothed$labels$title <- 'Smoothed Peak'
plot(original)
plot(smoothed)

Identifying Chromatographic Features

Chromatographic features are then identified using local maximum searching. The function get_peaks extracts all potential features from the chromatogram

peaks_found <- get_peaks(rt,smooth_int_opt)
plot_chrom(rt, smooth_int_opt, peak_info = peaks_found)

peaks_found %>%
  knitr::kable("html") %>%
  kable_styling(
  bootstrap_options = c("striped", "hover"),
  full_width = TRUE,
  position = 'center'
  ) %>%
  row_spec(4, bold = TRUE, background = "#ff6a6a")

From the table of results it is clear that the get_peaks function has successfully identified the apex of the peak (highlighted); aswell as a number of smaller chromatrographic features on either side.

As pickeR is primarily desgined to identify chromatographic features in SRM-MS data; then we can also use the get_chrom_max function to return only the retention time (Rt) of the major feature.

get_chrom_max(rt, smooth_int_opt)

Peak Picking

The peak_picker function is a wrapper which combines peak identification and merging. The methods implemented in get_peaks are not sophisticated enough to determine if neighbouring features (ie, peak shoulder) belong to the major feature. The peak_picker function identfies these minor features and combines them with the major feature identified.

By suppling a target retention time and window size (seconds) the extent of this searching and mering can be controlled.

target_rt <- get_chrom_max(rt, smooth_int_opt)

peaks <- peak_picker(rt, smooth_int_opt, target_rt, rt_window = 20)

peaks %>% knitr::kable("html") %>%
  kable_styling(
  bootstrap_options = c("striped", "hover"),
  full_width = TRUE,
  position = 'center'
  )

The identified peak can be visualised using the plot_area function;

plot_area(rt, smooth_int_opt, peaks)

wilsontom/pickeR documentation built on May 4, 2019, 7:43 p.m.