| Title: | Statistics and dIagnostic Graphs for HTS |
|---|---|
| Description: | SIGHTS is a suite of normalization methods, statistical tests, and diagnostic graphical tools for high throughput screening (HTS) assays. HTS assays use microtitre plates to screen large libraries of compounds for their biological, chemical, or biochemical activity. |
| Authors: | Elika Garg [aut, cre], Carl Murie [aut], Heydar Ensha [ctb], Robert Nadon [aut] |
| Maintainer: | Elika Garg <[email protected]> |
| License: | GPL-3 | file LICENSE |
| Version: | 1.31.0 |
| Built: | 2024-09-28 03:57:33 UTC |
| Source: | https://github.com/bioc/sights |
An example dataset containing High-Throughput Screening (HTS) output and experimental design information. See References for details.
data(ex_dataMatrix)data(ex_dataMatrix)
A data frame with 80 rows and 9 columns:
Wells. Plate well numbers for each sample
Rows. Plate row identifiers for each sample
Columns. Plate column identifiers for each sample
S1_R1. Screen 1 Replicate 1
S1_R2. Screen 1 Replicate 2
S1_R3. Screen 1 Replicate 3
S2_R1. Screen 2 Replicate 1
S2_R2. Screen 2 Replicate 2
S2_R3. Screen 2 Replicate 3
This example data matrix consists of 6 plates with 80 wells each. Although these are 96-well plates, only 80 wells in each plate contained the active compounds. Therefore, the subsequent data matrix for this package excludes the inactive wells.
The sights data format requires each plate matrix to be converted into a 1-dimensional vector. The plate wells in this vector should be arranged by row first. For example, this 3x3 plate matrix:
| Col 1 | Col 2 | Col 3 | |
| Row A | A1 | A2 | A3 |
| Row B | B1 | B2 | B3 |
| Row C | C1 | C2 | C3 |
can be converted into its vector form as:
| Row | Col | Data |
| A | 1 | A1 |
| A | 2 | A2 |
| A | 3 | A3 |
| B | 1 | B1 |
| B | 2 | B2 |
| B | 3 | B3 |
| C | 1 | C1 |
| C | 2 | C2 |
| C | 3 | C3 |
Here, number of columns in a plate is 3, and number of rows is 3 as well. Each such plate vector should form a column in the data matrix before application of sights functions. Only the active wells should be included in the data matrix; inactive wells containing mock/control compounds should be marked as NAs, or if they are in entire rows/columns they can be removed completely as in this example dataset and the arguments plateRows and plateCols modified accordingly.
Dataframe of 80 rows and 9 columns as explained in Format
CMBA Titration series 10uM Tyr samples. Murie et al. (2015). Improving detection of rare biological events in high-throughput screens. Journal of Biomolecular Screening, 20(2), 230-241.
## load dataset data(ex_dataMatrix) ## structure of dataset str(ex_dataMatrix) ## summary of dataset summary(ex_dataMatrix) ## See help pages of SIGHTS functions for examples of using this dataset## load dataset data(ex_dataMatrix) ## structure of dataset str(ex_dataMatrix) ## summary of dataset summary(ex_dataMatrix) ## See help pages of SIGHTS functions for examples of using this dataset
A published dataset containing High-Throughput Screening (HTS) output and experimental design information. See References for details.
data(inglese)data(inglese)
A data frame with 1280 rows and 45 columns:
Row. Plate row identifiers for each sample
Col. Plate column identifiers for each sample
Exp1R1. Screen 1 Replicate 1
Exp1R2. Screen 1 Replicate 2
Exp1R3. Screen 1 Replicate 3
Exp2R1. Screen 2 Replicate 1
Exp2R2. Screen 2 Replicate 2
Exp2R3. Screen 2 Replicate 3
... and so on until Exp14 totaling to 14 screens in triplicate.
Hits. Presence or absence of hits identified for each sample
Dataframe of 1280 rows and 45 columns as explained in Format
For information on how to arrange your dataset, please see (ex_dataMatrix)
Titration series samples. Inglese et al. (2006). Quantitative High-Throughput Screening: A Titration-Based Approach That Efficiently Identifies Biological Activities in Large Chemical Libraries. Proc. Natl. Acad. Sci. U. S. A., 103, 11473-11478.
## load dataset data(inglese) ## structure of dataset str(inglese) ## summary of dataset summary(inglese) ## See SIGHTS vignette for examples of using this dataset and its anlaysis## load dataset data(inglese) ## structure of dataset str(inglese) ## summary of dataset summary(inglese) ## See SIGHTS vignette for examples of using this dataset and its anlaysis
Apply loess normalization to data
normLoess(dataMatrix, plateRows, plateCols, dataRows = NULL, dataCols = NULL)normLoess(dataMatrix, plateRows, plateCols, dataRows = NULL, dataCols = NULL)
dataMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
plateRows, plateCols
|
Number of rows/columns in plate. |
dataRows, dataCols
|
Optional integer vector. Indicate which row/column numbers from the dataMatrix should be normalized. If NULL then all rows/columns from the dataMatrix are used. |
Loess normalization adjusts each well by the fitted row and column values generated by calculating the loess curve for each row and column.
Numeric matrix of normalized data in the same format as dataMatrix
For information on how to arrange your dataset for dataMatrix, please see (ex_dataMatrix)
Other normalization methods: normMedFil,
normRobZ, normR,
normSPAWN, normZ
## load dataset data(ex_dataMatrix) ## apply Loess method ex_normMatrix <- normLoess(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10)## load dataset data(ex_dataMatrix) ## apply Loess method ex_normMatrix <- normLoess(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10)
Apply median filter normalization to data
normMedFil(dataMatrix, plateRows, plateCols, dataRows = NULL, dataCols = NULL, seqFilter = TRUE)normMedFil(dataMatrix, plateRows, plateCols, dataRows = NULL, dataCols = NULL, seqFilter = TRUE)
dataMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
plateRows, plateCols
|
Number of rows/columns in plate. |
dataRows, dataCols
|
Optional integer vector. Indicate which row/column numbers from the dataMatrix should be normalized. If NULL then all rows/columns from the dataMatrix are used. |
seqFilter |
Optional logical. If TRUE apply initial row median filter then standard filter, else just apply standard filter. |
Median Filter normalization uses a two-step median filter process where each well is adjusted by the median score of a neighbouring group of wells [Bushway et al (2011)]. The first median filter uses a neighbour set based on the Manhattan distance to each well. The second median filter uses a neighbour set based on the proximity along each row or column.
Numeric matrix of normalized data in the same format as dataMatrix
For information on how to arrange your dataset for dataMatrix, please see (ex_dataMatrix)
Other normalization methods: normLoess,
normRobZ, normR,
normSPAWN, normZ
## load dataset data(ex_dataMatrix) ## apply standard median filter method ex_normMatrix <- normMedFil(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10, seqFilter = FALSE) ## apply initial row median filter then standard filter ex_normMatrix <- normMedFil(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10, seqFilter = TRUE)## load dataset data(ex_dataMatrix) ## apply standard median filter method ex_normMatrix <- normMedFil(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10, seqFilter = FALSE) ## apply initial row median filter then standard filter ex_normMatrix <- normMedFil(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10, seqFilter = TRUE)
Apply Robust Regression model separately to each plate
normR(dataMatrix, plateRows, plateCols, dataRows = NULL, dataCols = NULL)normR(dataMatrix, plateRows, plateCols, dataRows = NULL, dataCols = NULL)
dataMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
plateRows, plateCols
|
Number of rows/columns in plate. |
dataRows, dataCols
|
Optional integer vector. Indicate which row/column numbers from the dataMatrix should be normalized. If NULL then all rows/columns from the dataMatrix are used. |
R score normalization uses the robust regression method described by Wu et al (2008). Parameters are estimated through the rlm function. Data is pre-normalized by median normalization prior to applying the regression algorithm. R scores are the residuals produced by the model and rescaled by dividing with the standard deviation estimate from the regression function.
Numeric matrix of normalized data in the same format as dataMatrix
For information on how to arrange your dataset for dataMatrix, please see (ex_dataMatrix)
Other normalization methods: normLoess,
normMedFil, normRobZ,
normSPAWN, normZ
## load dataset data(ex_dataMatrix) ## apply R score ex_normMatrix <- normR(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10)## load dataset data(ex_dataMatrix) ## apply R score ex_normMatrix <- normR(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10)
Apply robust Z score to data
normRobZ(dataMatrix, dataRows = NULL, dataCols = NULL)normRobZ(dataMatrix, dataRows = NULL, dataCols = NULL)
dataMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
dataRows, dataCols
|
Optional integer vector. Indicate which row/column numbers from the dataMatrix should be normalized. If NULL then all rows/columns from the dataMatrix are used. |
Robust Z score normalization subtracts the median of the raw well intensities of a given plate from the signal intensity of a given compound and divides it by the median absolute deviation of the raw well intensities of that plate.
Numeric matrix of normalized data in the same format as dataMatrix
For information on how to arrange your dataset for dataMatrix, please see (ex_dataMatrix)
Other normalization methods: normLoess,
normMedFil, normR,
normSPAWN, normZ
## load dataset data(ex_dataMatrix) ## apply robust Z score ex_normMatrix <- normRobZ(dataMatrix = ex_dataMatrix, dataCols = 5:10)## load dataset data(ex_dataMatrix) ## apply robust Z score ex_normMatrix <- normRobZ(dataMatrix = ex_dataMatrix, dataCols = 5:10)
Apply any of the available SIGHTS normalization methods
normSights(normMethod, dataMatrix, plateRows, plateCols, dataRows = NULL, dataCols = NULL, trimFactor = 0.2, wellCorrection = FALSE, biasMatrix = NULL, biasCols = NULL, seqFilter = TRUE)normSights(normMethod, dataMatrix, plateRows, plateCols, dataRows = NULL, dataCols = NULL, trimFactor = 0.2, wellCorrection = FALSE, biasMatrix = NULL, biasCols = NULL, seqFilter = TRUE)
normMethod |
Normalization method name from SIGHTS ('Z', 'RobZ', 'R', 'SPAWN', 'Loess', or 'MedFil') |
dataMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
plateRows, plateCols
|
Number of rows/columns in plate. Applies to normMethods 'R', 'SPAWN', 'Loess', and 'MedFil'. |
dataRows, dataCols
|
Optional integer vector. Indicate which row/column numbers from the dataMatrix should be normalized. If NULL then all rows/columns from the dataMatrix are used. |
trimFactor |
Optional trim value to be used in trimmed mean polish. It should be between 0 and 0.5. Default is 0.2. Applies to normMethod 'SPAWN'. |
wellCorrection |
Optional logical. If TRUE then individual wells are corrected based on spatial bias. Applies to normMethod 'SPAWN'. |
biasMatrix |
Optional data frame or numeric matrix, in the same format as dataMatrix and with the same plateRows and plateCols specifications. If NULL then normalized data is used as bias template. Applies to normMethod 'SPAWN'. |
biasCols |
Optional integer vector. Indicate which column numbers from biasMatrix or normalized dataMatrix (subset of dataCols) should be used to calculate bias template. Control plates or selection of dataMatrix plates to be used for well correction. If NULL then all plates of biasMatrix or normalized dataMatrix are used. Applies to normMethod 'SPAWN'. |
seqFilter |
Optional logical. If TRUE apply initial row median filter then standard filter, else just apply standard filter. Applies to normMethod 'MedFil'. |
One of the following SIGHTS normalization methods may be chosen: normZ, normRobZ, normR, normSPAWN, normLoess, or normMedFil. See their individual help pages for more details.
Numeric matrix of normalized data in the same format as dataMatrix
For information on how to arrange your dataset for dataMatrix, please see (ex_dataMatrix)
Other SIGHTS functions: plotSights,
statSights
## load dataset data(ex_dataMatrix) ## choose a normalization method and provide relevant information ex_normMatrix <- normSights(dataMatrix = ex_dataMatrix, dataCols = 5:10, normMethod = 'RobZ')## load dataset data(ex_dataMatrix) ## choose a normalization method and provide relevant information ex_normMatrix <- normSights(dataMatrix = ex_dataMatrix, dataCols = 5:10, normMethod = 'RobZ')
Apply trimmed mean polish to data
normSPAWN(dataMatrix, plateRows, plateCols, dataRows = NULL, dataCols = NULL, trimFactor = 0.2, wellCorrection = FALSE, biasMatrix = NULL, biasCols = NULL)normSPAWN(dataMatrix, plateRows, plateCols, dataRows = NULL, dataCols = NULL, trimFactor = 0.2, wellCorrection = FALSE, biasMatrix = NULL, biasCols = NULL)
dataMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
plateRows, plateCols
|
Number of rows/columns in plate. |
dataRows, dataCols
|
Optional integer vector. Indicate which row/column numbers from the dataMatrix should be normalized. If NULL then all rows/columns from the dataMatrix are used. |
trimFactor |
Optional trim value to be used in trimmed polish. It should be between 0 and 0.5. Default is 0.2. |
wellCorrection |
Optional logical. If TRUE then individual wells are corrected based on spatial bias. |
biasMatrix |
Optional data frame or numeric matrix, in the same format as dataMatrix and with the same plateRows and plateCols specifications. If NULL then normalized data is used as bias template. |
biasCols |
Optional integer vector. Indicate which column numbers from biasMatrix or normalized dataMatrix (subset of dataCols) should be used to calculate bias template. Control plates or selection of dataMatrix plates to be used for well correction. If NULL then all plates of biasMatrix or normalized dataMatrix are used. |
Spatial Polish And Well Normalization (SPAWN) uses a trimmed mean polish on individual plates to remove row and column effects. Data from each well location on each plate are initially fitted to the same model as the R score. Model parameters are estimated with an iterative polish technique but with a trimmed mean, rather than a median, as a measure of central tendency for row and column effects. The residuals are rescaled by dividing by the median average deviation of their respective plates. Well correction uses a bias template, which can either be the normalized plates themselves or be supplied externally (and SPAWN normalized before application). At each well location of this bias template, a median of all plates is calculated and subtracted from the normalized plates, thus correcting for well location bias.
Numeric matrix of normalized data in the same format as dataMatrix
For information on how to arrange your dataset for dataMatrix, please see (ex_dataMatrix)
Other normalization methods: normLoess,
normMedFil, normRobZ,
normR, normZ
## load dataset data(ex_dataMatrix) ## apply SPAWN method with default trim factor and without well correction ex_normMatrix <- normSPAWN(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10, trimFactor = 0.2) ## apply SPAWN method with default trim factor and with well correction ex_normMatrix <- normSPAWN(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10, trimFactor = 0.2, wellCorrection = TRUE)## load dataset data(ex_dataMatrix) ## apply SPAWN method with default trim factor and without well correction ex_normMatrix <- normSPAWN(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10, trimFactor = 0.2) ## apply SPAWN method with default trim factor and with well correction ex_normMatrix <- normSPAWN(dataMatrix = ex_dataMatrix, dataCols = 5:10, plateRows = 8, plateCols = 10, trimFactor = 0.2, wellCorrection = TRUE)
Apply Z score to data
normZ(dataMatrix, dataRows = NULL, dataCols = NULL)normZ(dataMatrix, dataRows = NULL, dataCols = NULL)
dataMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
dataRows, dataCols
|
Optional integer vector. Indicate which row/column numbers from the dataMatrix should be normalized. If NULL then all rows/columns from the dataMatrix are used. |
Z score normalization subtracts the mean of the raw well intensities of a given plate from the signal intensity of a given compound and divides it by the standard deviation of the raw well intensities of that plate.
Numeric matrix of normalized data in the same format as dataMatrix
For information on how to arrange your dataset for dataMatrix, please see (ex_dataMatrix)
Other normalization methods: normLoess,
normMedFil, normRobZ,
normR, normSPAWN
## load dataset data(ex_dataMatrix) ## apply Z score ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10)## load dataset data(ex_dataMatrix) ## apply Z score ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10)
Plot a three-dimensional plot for each plate
plot3d(plotMatrix, plateRows, plateCols, plotRows = NULL, plotCols = NULL, plotName = NULL)plot3d(plotMatrix, plateRows, plateCols, plotRows = NULL, plotCols = NULL, plotName = NULL)
plotMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
plateRows, plateCols
|
Number of rows/columns in plate. |
plotRows, plotCols
|
Optional integer vector. Indicate which row/column numbers from the plotMatrix should be plotted. If NULL then all rows/columns from the plotMatrix are used. |
plotName |
Optional. Name of plotMatrix for plot title. |
3d plots can be used to assess the existence of spatial bias on a plate by plate basis. Spatial bias can be visually subtle, however, and sometimes difficult to detect with 3d plots. Auto-correlation plots (plotAutoco) can circumvent this problem.
List of lattice objects
Other graphical devices: plotAutoco,
plotBox, plotHeatmap,
plotHist, plotIGFit,
plotScatter
## load dataset data(ex_dataMatrix) ## plot raw data plot3d(plotMatrix = ex_dataMatrix, plotCols = 5:10, plotName = 'Example', plateRows = 8, plateCols = 10) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## plot normalized data plot3d(plotMatrix = ex_normMatrix, plotName = 'Example', plateRows = 8, plateCols = 10)## load dataset data(ex_dataMatrix) ## plot raw data plot3d(plotMatrix = ex_dataMatrix, plotCols = 5:10, plotName = 'Example', plateRows = 8, plateCols = 10) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## plot normalized data plot3d(plotMatrix = ex_normMatrix, plotName = 'Example', plateRows = 8, plateCols = 10)
Plot auto-correlation for each plate
plotAutoco(plotMatrix, plateRows, plateCols, plotRows = NULL, plotCols = NULL, plotName = NULL, plotSep = TRUE, ...)plotAutoco(plotMatrix, plateRows, plateCols, plotRows = NULL, plotCols = NULL, plotName = NULL, plotSep = TRUE, ...)
plotMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
plateRows, plateCols
|
Number of rows/columns in plate. |
plotRows, plotCols
|
Optional integer vector. Indicate which row/column numbers from the plotMatrix should be plotted. If NULL then all rows/columns from the plotMatrix are used. |
plotName |
Optional. Name of plotMatrix for plot title. |
plotSep |
Optional logical. Should plots be presented in separate windows? Default is TRUE. |
... |
Optional. Additional parameters passed to |
Auto-correlation plots can be used to identify spatial bias. Non-zero auto-correlations indicate within-plate bias, namely that proximal wells within-plates are correlated and that the measured intensity of a feature depends partially on its well location in the plate. Cyclical patterns of auto-correlation, in particular indicate within-plate spatial bias. Normalization methods that produce auto-correlations close to zero indicate the removal of spatial bias.
Modifiable ggplot2 object or list of objects
Other graphical devices: plot3d,
plotBox, plotHeatmap,
plotHist, plotIGFit,
plotScatter
## load dataset data(ex_dataMatrix) ## plot raw data plotAutoco(plotMatrix = ex_dataMatrix, plateRows = 8, plateCols = 10, plotCols = 5:10, plotName = 'Example') ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## plot normalized data plotAutoco(plotMatrix = ex_normMatrix, plotName = 'Example', plateRows = 8, plateCols = 10, plotSep = FALSE)## load dataset data(ex_dataMatrix) ## plot raw data plotAutoco(plotMatrix = ex_dataMatrix, plateRows = 8, plateCols = 10, plotCols = 5:10, plotName = 'Example') ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## plot normalized data plotAutoco(plotMatrix = ex_normMatrix, plotName = 'Example', plateRows = 8, plateCols = 10, plotSep = FALSE)
Construct an ordered boxplot for each plate
plotBox(plotMatrix, plotRows = NULL, plotCols = NULL, plotName = NULL, repIndex = NULL, plotSep = TRUE, ...)plotBox(plotMatrix, plotRows = NULL, plotCols = NULL, plotName = NULL, repIndex = NULL, plotSep = TRUE, ...)
plotMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
plotRows, plotCols
|
Optional integer vector. Indicate which row/column numbers from the plotMatrix should be plotted. If NULL then all rows/columns from the plotMatrix are used. |
plotName |
Optional. Name of plotMatrix for plot title. |
repIndex |
Optional. Vector of labels indicating replicate group. Each index in the vector matches the corresponding column of plotMatrix. If NULL then all plates are plotted together without grouping. |
plotSep |
Optional logical. Should plots of different replicate groups be presented in separate windows? Default is TRUE. Does not apply if repIndex is NULL. |
... |
Optional. Additional parameters passed to |
Box plots can be used to identify scaling shifts among replicates and view the general distribution of data among all plates.
Modifiable ggplot2 object or list of objects
Other graphical devices: plot3d,
plotAutoco, plotHeatmap,
plotHist, plotIGFit,
plotScatter
## load dataset data(ex_dataMatrix) ## plot raw data plotBox(plotMatrix = ex_dataMatrix, repIndex = c(1,1,1,2,2,2), plotCols = 5:10, plotName = 'Example') ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## plot normalized data plotBox(plotMatrix = ex_normMatrix, repIndex = c(1,1,1,2,2,2), plotName = 'Example')## load dataset data(ex_dataMatrix) ## plot raw data plotBox(plotMatrix = ex_dataMatrix, repIndex = c(1,1,1,2,2,2), plotCols = 5:10, plotName = 'Example') ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## plot normalized data plotBox(plotMatrix = ex_normMatrix, repIndex = c(1,1,1,2,2,2), plotName = 'Example')
Plot heat map for each plate
plotHeatmap(plotMatrix, plateRows, plateCols, plotRows = NULL, plotCols = NULL, plotName = NULL, plotSep = TRUE, ...)plotHeatmap(plotMatrix, plateRows, plateCols, plotRows = NULL, plotCols = NULL, plotName = NULL, plotSep = TRUE, ...)
plotMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
plateRows, plateCols
|
Number of rows/columns in plate. |
plotRows, plotCols
|
Optional integer vector. Indicate which row/column numbers from the plotMatrix should be plotted. If NULL then all rows/columns from the plotMatrix are used. |
plotName |
Optional. Name of plotMatrix for plot title. |
plotSep |
Optional logical. Should plots be presented in separate windows? Default is TRUE. |
... |
Optional. Additional parameters passed to |
Heat maps can be used to assess the existence of spatial bias on a plate by plate basis. Spatial bias can be visually subtle, however, and sometimes difficult to detect with heat maps. Auto-correlation plots (plotAutoco) can circumvent this problem.
Modifiable ggplot2 object or list of objects
Other graphical devices: plot3d,
plotAutoco, plotBox,
plotHist, plotIGFit,
plotScatter
## load dataset data(ex_dataMatrix) ## plot raw data with graphs separated plotHeatmap(plotMatrix = ex_dataMatrix, plotCols = 5:10, plotName = 'Example', plateRows = 8, plateCols = 10) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## plot normalized data with graphs together plotHeatmap(plotMatrix = ex_normMatrix, plotName = 'Example', plateRows = 8, plateCols = 10, plotSep = FALSE)## load dataset data(ex_dataMatrix) ## plot raw data with graphs separated plotHeatmap(plotMatrix = ex_dataMatrix, plotCols = 5:10, plotName = 'Example', plateRows = 8, plateCols = 10) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## plot normalized data with graphs together plotHeatmap(plotMatrix = ex_normMatrix, plotName = 'Example', plateRows = 8, plateCols = 10, plotSep = FALSE)
Plot histogram of p-values or q-values for each plate or all plates together
plotHist(plotMatrix, plotRows = NULL, plotCols = NULL, plotAll = FALSE, plotSep = TRUE, plotName = NULL, colNames = NULL, ...)plotHist(plotMatrix, plotRows = NULL, plotCols = NULL, plotAll = FALSE, plotSep = TRUE, plotName = NULL, colNames = NULL, ...)
plotMatrix |
Data frame or numeric matrix consisting only of p-values or q-values. Columns are samples, and rows are plate wells. |
plotRows, plotCols
|
Optional integer vector. Indicate which row/column numbers from the plotMatrix should be plotted. If NULL then all rows/columns from the plotMatrix are used. |
plotAll |
Optional logical. Should all p-values or q-values be plotted together? Default is FALSE. |
plotSep |
Optional logical. If plotAll is FALSE, should plots be presented in separate windows? Default is TRUE. |
plotName |
Optional. Name of plotMatrix for plot title. |
colNames |
Optional. If plotAll is FALSE, names of plotCols for plot titles. |
... |
Optional. Additional parameters passed to |
Histograms can be used to compare actual to expected p-value distributions obtained from statistical tests of replicated features. In the presence of rare biological events, the p-value distribution should be approximately uniformly distributed with somewhat more small p-values. Deviations from these patterns indicate that the activity measurements are incorrect and/or that the statistical model is incorrectly specified.
Modifiable ggplot2 object or list of objects
If using output from statT, statRVM, statFDR or statSights, please only select the plotCols corresponding to p-value and/or q-value columns, i.e., every 5th and/or 6th column in that output. Also, the x-axis label is derived from these column names indicating either 'p-values' or 'q-values'.
Other graphical devices: plot3d,
plotAutoco, plotBox,
plotHeatmap, plotIGFit,
plotScatter
## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## apply any test to normalized data and store in new variable ex_testMatrix <- statRVM(normMatrix = ex_normMatrix, repIndex = c(1,1,1,2,2,2)) ## plot p-value data by selecting the p-value columns from test result matrix plotHist(plotMatrix = ex_testMatrix, plotCols = c(5,10), plotName = 'Example', colNames = c('Set_A', 'Set_B'))## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## apply any test to normalized data and store in new variable ex_testMatrix <- statRVM(normMatrix = ex_normMatrix, repIndex = c(1,1,1,2,2,2)) ## plot p-value data by selecting the p-value columns from test result matrix plotHist(plotMatrix = ex_testMatrix, plotCols = c(5,10), plotName = 'Example', colNames = c('Set_A', 'Set_B'))
Plot an inverse gamma fit plot for all plates together
plotIGFit(plotMatrix, repIndex, plotRows = NULL, plotCols = NULL, plotName = NULL, ...)plotIGFit(plotMatrix, repIndex, plotRows = NULL, plotCols = NULL, plotName = NULL, ...)
plotMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
repIndex |
Optional. Vector of labels indicating replicate group. Each index in the vector matches the corresponding column of plotMatrix. If NULL then all plates are plotted together without grouping. |
plotRows, plotCols
|
Optional integer vector. Indicate which row/column numbers from the plotMatrix should be plotted. If NULL then all rows/columns from the plotMatrix are used. |
plotName |
Optional. Name of plotMatrix for plot title. |
... |
Optional. Additional parameters passed to |
Inverse gamma fit plot can be used to check if RVM test (statRVM) assumptions are valid and it can be applied to the data.
Modifiable ggplot2 object
Other graphical devices: plot3d,
plotAutoco, plotBox,
plotHeatmap, plotHist,
plotScatter
## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(dataMatrix = ex_dataMatrix, dataCols = 5:10, normMethod = 'normZ') ## plot normalized data plotIGFit(plotMatrix = ex_normMatrix, repIndex = c(1,1,1,2,2,2), plotName = 'Example')## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(dataMatrix = ex_dataMatrix, dataCols = 5:10, normMethod = 'normZ') ## plot normalized data plotIGFit(plotMatrix = ex_normMatrix, repIndex = c(1,1,1,2,2,2), plotName = 'Example')
Construct a scatter plot of all pairwise combinations of replicates
plotScatter(plotMatrix, repIndex, plotRows = NULL, plotCols = NULL, plotName = NULL, ...)plotScatter(plotMatrix, repIndex, plotRows = NULL, plotCols = NULL, plotName = NULL, ...)
plotMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. |
repIndex |
Optional. Vector of labels indicating replicate group. Each index in the vector matches the corresponding column of plotMatrix. If NULL then all plates are plotted together without grouping. |
plotRows, plotCols
|
Optional integer vector. Indicate which row/column numbers from the plotMatrix should be plotted. If NULL then all rows/columns from the plotMatrix are used. |
plotName |
Optional. Name of plotMatrix for plot title. |
... |
Optional. Additional parameters passed to |
Scatter plots with robust regression lines of replicate plates can reveal a kind of bias which acts independently of within-plate biases and which cannot be detected by heat maps (plotHeatmap) or auto-correlation plots (plotAutoco). A mixture of active and inactive features should produce a zero-correlation flat regression line within most of the range and a positively sloped line within the active range(s) at the extreme(s) of the distribution.
List of modifiable ggplot2 objects
Other graphical devices: plot3d,
plotAutoco, plotBox,
plotHeatmap, plotHist,
plotIGFit
## load dataset data(ex_dataMatrix) ## plot raw data plotScatter(plotMatrix = ex_dataMatrix, repIndex = c(1,1,1), plotCols = 5:7, plotName = 'Example') ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## plot normalized data plotScatter(plotMatrix = ex_normMatrix, repIndex = c(1,1,1), plotCols = 1:3, plotName = 'Example')## load dataset data(ex_dataMatrix) ## plot raw data plotScatter(plotMatrix = ex_dataMatrix, repIndex = c(1,1,1), plotCols = 5:7, plotName = 'Example') ## normalize data matrix using any method and store in new variable ex_normMatrix <- normZ(dataMatrix = ex_dataMatrix, dataCols = 5:10) ## plot normalized data plotScatter(plotMatrix = ex_normMatrix, repIndex = c(1,1,1), plotCols = 1:3, plotName = 'Example')
Apply any of the available SIGHTS graphical devices
plotSights(plotMethod, plotMatrix, plateRows, plateCols, repIndex = NULL, plotRows = NULL, plotCols = NULL, plotName = NULL, plotSep = TRUE, plotAll = FALSE, colNames = NULL, ...)plotSights(plotMethod, plotMatrix, plateRows, plateCols, repIndex = NULL, plotRows = NULL, plotCols = NULL, plotName = NULL, plotSep = TRUE, plotAll = FALSE, colNames = NULL, ...)
plotMethod |
Plotting method name from SIGHTS ('3d', 'Autoco', 'Box', 'Heatmap', 'Hist', 'IGFit', or 'Scatter'). |
plotMatrix |
Data frame or numeric matrix. Columns are plates, and rows are plate wells. For plotMethod 'Hist', this is a p-value matrix with each column a single sample. |
plateRows, plateCols
|
Number of rows/columns in plate. Applies to plotMethods '3d', 'Autoco' and 'Heatmap'. |
repIndex |
Vector of labels indicating replicate group. Each index in the vector matches the corresponding column of plotMatrix. Applies to plotMethods 'Box', 'Scatter' and 'IGFit'. |
plotRows, plotCols
|
Optional integer vector. Indicate which row/column numbers from the plotMatrix should be plotted. If NULL then all rows/columns from the plotMatrix are used. |
plotName |
Optional. Name of plotMatrix for plot title. |
plotSep |
Optional logical. Should plots be presented in separate windows? Default is TRUE. Applies to plotMethods 'Autoco', 'Box', 'Hist' and 'Heatmap'. For 'Box', each replicate group is presented in a separate window and it only applies if repIndex is not NULL. |
plotAll |
Optional logical. Should all p-values be plotted together? Default is FALSE. Applies to plotMethod 'Hist'. |
colNames |
Optional. Names of plotCols for plot title. Applies to plotMethod 'Hist'. |
... |
Optional. Additional parameters passed to |
One of the following SIGHTS graphical devices may be chosen: plot3d, plotAutoco, plotBox, plotHeatmap, plotHist, plotIGFit, or plotScatter. See their individual help pages for more details.
List of lattice objects for 'plot3d'. Modifiable ggplot2 object or list of objects for all others.
Other SIGHTS functions: normSights,
statSights
## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(normMethod = 'RobZ', dataMatrix = ex_dataMatrix, dataCols = 5:10, wellCorrection = TRUE) ## choose a graphical device and provide relevant information plotSights(plotMethod = 'Autoco', plotMatrix = ex_normMatrix, plotName = 'Example', plateRows = 8, plateCols = 10)## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(normMethod = 'RobZ', dataMatrix = ex_dataMatrix, dataCols = 5:10, wellCorrection = TRUE) ## choose a graphical device and provide relevant information plotSights(plotMethod = 'Autoco', plotMatrix = ex_normMatrix, plotName = 'Example', plateRows = 8, plateCols = 10)
Apply Storey's FDR control to p-values
statFDR(testMatrix, ctrlMethod = "smoother", ...)statFDR(testMatrix, ctrlMethod = "smoother", ...)
testMatrix |
Data frame or numeric matrix consisting of output from |
ctrlMethod |
Optional. Method to use either 'smoother' or 'bootstrap' to estimate null. Default is 'smoother'. |
... |
Optional. Additional parameters passed to |
False Discovery Rate procedure is used to control the proportion of false positives in the results. This is an implementation of the positive false discovery (pFDR) procedure of the qvalue function.
A matrix of parameters for each replicate group is returned:
T-statistic or RVM T-statistic |
Value of the t-statistic. |
Mean_Difference |
Difference between the calculated and the true mean. |
Standard_Error |
Standard error of the difference between means. |
Degrees_Of_Freedom |
Degrees of freedom for the t-statistic. |
P-value |
P-value for the t-test. |
q-value |
FDR q-value for the P-value. |
Please install the package 'qvalue' from Bioconductor, if not already installed.
Other statistical methods: statRVM,
statT
## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(dataMatrix = ex_dataMatrix, dataCols = 5:10, normMethod = 'normZ') ## test normalized data matrix using either the RVM or T test and store in new variable ex_testMatrix <- statT(normMatrix = ex_normMatrix, trueMean = 0, repIndex = c(1,1,1,2,2,2)) ## apply FDR control to test matrix with bootstrap control method ex_ctrlMatrix <- statFDR(testMatrix = ex_testMatrix, ctrlMethod = 'bootstrap')## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(dataMatrix = ex_dataMatrix, dataCols = 5:10, normMethod = 'normZ') ## test normalized data matrix using either the RVM or T test and store in new variable ex_testMatrix <- statT(normMatrix = ex_normMatrix, trueMean = 0, repIndex = c(1,1,1,2,2,2)) ## apply FDR control to test matrix with bootstrap control method ex_ctrlMatrix <- statFDR(testMatrix = ex_testMatrix, ctrlMethod = 'bootstrap')
Apply one-sample RVM t-test separately to each plate
statRVM(normMatrix, repIndex, normRows = NULL, normCols = NULL, testSide = "two.sided")statRVM(normMatrix, repIndex, normRows = NULL, normCols = NULL, testSide = "two.sided")
normMatrix |
Data frame or numeric matrix of normalized data. Columns are plates, and rows are plate wells. |
repIndex |
Integer vector indicating replicates in normMatrix. Which plates are replicates of each other? Provide the same number for plates belonging to a replicate group. Each index in the vector matches the corresponding column of normMatrix. |
normRows, normCols
|
Optional integer vector. Indicate which row/column numbers from the normMatrix should be tested. If NULL then all rows/columns from the normMatrix are used. |
testSide |
Optional. Type of t-test: 'two.sided', 'less', or 'greater'. Default is 'two.sided'. |
Random Variance Model one-sample t-test is applied to the normalized data. RVM assumes that the across replicate variances are distributed according to an inverse gamma distribution. This can be checked by using the plotIGFit function.
A matrix of parameters for each replicate group is returned:
RVM T-statistic |
Value of the RVM t-statistic. |
Mean_Difference |
Difference between the calculated and the true mean. |
Standard_Error |
Standard error of the difference between means. |
Degrees_Of_Freedom |
Degrees of freedom for the t-statistic. |
P-value |
P-value for the RVM test. |
Other statistical methods: statFDR,
statT
## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(dataMatrix = ex_dataMatrix, dataCols = 5:10, normMethod = 'normZ') ## apply RVM test to normalized data matrix and get the p-values ex_testMatrix <- statRVM(normMatrix = ex_normMatrix, repIndex = c(1,1,1,2,2,2))## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(dataMatrix = ex_dataMatrix, dataCols = 5:10, normMethod = 'normZ') ## apply RVM test to normalized data matrix and get the p-values ex_testMatrix <- statRVM(normMatrix = ex_normMatrix, repIndex = c(1,1,1,2,2,2))
Apply any of the available SIGHTS statistical methods
statSights(statMethod, normMatrix, repIndex, normRows = NULL, normCols = NULL, ctrlMethod = NULL, testSide = "two.sided", trueMean = 0, ...)statSights(statMethod, normMatrix, repIndex, normRows = NULL, normCols = NULL, ctrlMethod = NULL, testSide = "two.sided", trueMean = 0, ...)
statMethod |
Statistical testing method to use either 'T' or 'RVM'. |
normMatrix |
Data frame or numeric matrix of normalized data. Columns are plates, and rows are plate wells. |
repIndex |
Integer vector indicating replicates in normMatrix. Which plates are replicates of each other? Provide the same number for plates belonging to a replicate group. Each index in the vector matches the corresponding column of normMatrix. |
normRows, normCols
|
Optional integer vector. Indicate which row/column numbers from the normMatrix should be tested. If NULL then all rows/columns from the normMatrix are used. |
ctrlMethod |
Optional. FDR method to use either 'smoother' or 'bootstrap' to estimate null. Default is NULL, which does not apply FDR control to the statistical testing output. |
testSide |
Optional. Type of t-test: 'two.sided', 'less', or 'greater'. Default is 'two.sided'. |
trueMean |
Optional. Number indicating true value of mean. Applies to statMethod 'T'. Default is 0. |
... |
Optional. Additional parameters passed to |
One of the two SIGHTS statistical testing methods may be chosen: statT or statRVM, and FDR control may be applied by statFDR. See their individual help pages for more details.
A matrix of parameters for each replicate group including p-values and q-values, if FDR control is applied.
Other SIGHTS functions: normSights,
plotSights
## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(normMethod = 'RobZ', dataMatrix = ex_dataMatrix, dataCols = 5:10, wellCorrection = TRUE) ## choose a statistical testing method, indicate FDR control ## and provide relevant information ex_statMatrix <- statSights(normMatrix = ex_normMatrix, statMethod = 'RVM', ctrlMethod = 'smoother', repIndex = c(1,1,1,2,2,2))## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(normMethod = 'RobZ', dataMatrix = ex_dataMatrix, dataCols = 5:10, wellCorrection = TRUE) ## choose a statistical testing method, indicate FDR control ## and provide relevant information ex_statMatrix <- statSights(normMatrix = ex_normMatrix, statMethod = 'RVM', ctrlMethod = 'smoother', repIndex = c(1,1,1,2,2,2))
Apply one-sample t-test separately to each plate
statT(normMatrix, repIndex, normRows = NULL, normCols = NULL, testSide = "two.sided", trueMean = 0)statT(normMatrix, repIndex, normRows = NULL, normCols = NULL, testSide = "two.sided", trueMean = 0)
normMatrix |
Data frame or numeric matrix of normalized data. Columns are plates, and rows are plate wells. |
repIndex |
Integer vector indicating replicates in normMatrix. Which plates are replicates of each other? Provide the same number for plates belonging to a replicate group. Each index in the vector matches the corresponding column of normMatrix. |
normRows, normCols
|
Optional integer vector. Indicate which row/column numbers from the normMatrix should be tested. If NULL then all rows/columns from the normMatrix are used. |
testSide |
Optional. Type of t-test: 'two.sided', 'less', or 'greater'. Default is 'two.sided'. |
trueMean |
Optional. Number indicating true value of mean. Default is 0. |
Standard one-sample t-test is applied to the normalized data.
A matrix of parameters for each replicate group is returned:
T-statistic |
Value of the t-statistic. |
Mean_Difference |
Difference between the calculated and the true mean. |
Standard_Error |
Standard error of the difference between means. |
Degrees_Of_Freedom |
Degrees of freedom for the t-statistic. |
P-value |
P-value for the t-test. |
Other statistical methods: statFDR,
statRVM
## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(dataMatrix = ex_dataMatrix, dataCols = 5:10, normMethod = 'normZ') ## apply T test to normalized data matrix and get the p-values ex_testMatrix <- statT(normMatrix = ex_normMatrix, trueMean = 0, repIndex = c(1,1,1,2,2,2))## load dataset data(ex_dataMatrix) ## normalize data matrix using any method and store in new variable ex_normMatrix <- normSights(dataMatrix = ex_dataMatrix, dataCols = 5:10, normMethod = 'normZ') ## apply T test to normalized data matrix and get the p-values ex_testMatrix <- statT(normMatrix = ex_normMatrix, trueMean = 0, repIndex = c(1,1,1,2,2,2))