Detecting periodic patterns
in unevenly spaced gene expression time series
using Lomb-Scargle periodograms

Home    Paper    Supplementary Information    R Code    Feedback

A.  Numerical experiments
B.  Exploratory analysis
C.  Results

1Box plot of raw data in Bozdech "Complete" Dataset

Also see notes about Bozdech03 Raw Data for a summary of the various datasets Bozdech used:  Complete, Quality Control, Overview.

Chart created as part of function ReadDataset in Bozdech03.R (called by LoadPlasmodiumData).

2Extreme points rejected  Obvious, single-point extreme points (> 3 standard deviations from the mean) were replaced with "missing values."  Negative ratios were replaced with missing values.

Because these extreme missing values were quite "high frequency" the Lomb-Scargle analysis was not sensitive to whether they were rejected or not.

3Box plot of raw data after extreme points removed

Function FixData in Bozdech03.R replaces extreme values with R's "na" and plots this chart.

4Box plot of log2 of raw data. 

Function PlotLog2RatioBoxPlot in Bozdech03.R (called by ProcessBozdech03).

5Frequency of missing values by time period
Column summary of missing data.  Note all points for time periods 23 and 29 are missing.

Function ShowMissingValuesByTimePeriod in Bozdech03.R

6Histogram shows number of oligos with various numbers of missing values   Row summary of missing data.

Function ShowMissingValueHistogramByOligo in Bozdech03.R.

7"Image matrix" of "complete" dataset

Note:  white space shows locations of missing values in dataset. Two vertical lines are visible for the missing values for time periods 23 and 29.

Function ShowLogRatioImage in Bozdech03.R.

A.  Numerical experiments  |  C.  Results

Updated
22 Sept 2005