readDataFile {mixlow}R Documentation

Read a formatted file of concentration-response data

Description

Reads a formatted data file that contains concentration-response information.

Usage

readDataFile(filename, excludeWells)

Arguments

filename A string designating the data file to be read.
excludeWells An optional list that specifies rows and columns that should be skipped in the analysis. Paired components of the list are:
row
row to skip
col
column to skip

Details

The data file read by this function must be formatted according to a specific structure. See the example file mixlowExampleData.txt for use as a template. The data file is tab-delimited pure ASCII (without enclosing quotations) and the first entry on each line is a line label. The first eight lines contain general information about the experiments. Of these lines, only the entries conc_units, rows, and cols are required. The remainder of the data file is split into blocks, one block for each tray. For each block, the required entries are tray_label, cell_line, drug_name_short, composition, as well as the entries for conc (concentration), label, and resp (response) for each row and column of a tray. Even if the optional entries are left blank, the line labels must still be included (including the “space” labels).

The “ray” design should be used for each experiment. In this design, various dilutions of a drug or mixture are tested in the wells. If a mixture is being tested, a fixed ratio between component drugs occurs at every concentration tested. A tray should contain data for only a single drug or mixture. Replicate trays should be analyzed for each drug or mixture (for a given cell line). At least three replicate trays are recommended. However, if only one tray is available for a drug, doNlme will duplicate that tray (to make two replicates) so that analysis can be done in “experimental” mode. Tray labels can be alphanumeric, but avoid any characters other than digits and letters (the underscore can be used). The same is true for the abbreviated drug names that are required.

Each tray should contain wells used as treatment controls (i.e., those that receive cells but no drug). In addition, trays should contain wells used as optical controls (called “blanks” here). These can be of two types: 1) wells that contain only media, and 2) wells at each drug concentration that contain media and drug. The first type is referred to as “bbt”, or blanks-by-tray, and the second type is referred to as “bbc”, or blanks-by-concentration. The use of “bbc” is recommended when concentration-dependent responses can be induced by the drug alone (e.g., autofluorescence). For type “bbt”, raw responses are adjusted by subtracting the mean response of all “bbt” wells in a tray. For type “bbc”, a 4th degree polynomial is fit to the “bbc” responses and predicted values for each concentration are subtracted from treatment well responses.

In many cases, a reasonable arrangement for a 96-well tray in a cytotoxicity experiment is to use the last two rows of every column as “bbc” blanks, use the first six rows of the first two columns as treatment controls, and use the first six rows of the third to twelfth columns as treatment wells, with each column receiving a different drug concentration.

When conducting large simulations, it is recommended that many small files be read from disk rather than one large file. The time required to read large files can be prohibitive.

Value

A list of class trayData with components:

concentrationResponse Data frame containing concentration- response information
drugRatios Data frame containing summary information for each tray

Author(s)

John Boik jcboik@stanford.edu

See Also

prepareData, summarizeData

Examples

dataFile = system.file(package="mixlow", "exdata/mixlowExampleData.txt")
excludeWells= list(row= c(1,2), col= c(1,1))
trayData <- readDataFile(filename=dataFile, excludeWells=excludeWells)


[Package mixlow version 0.01.1 Index]