Configuration files specifying the design of a compound screen on 84 NSCLC cell lines

Share:

Description

The experimental setup in a compound screen on 84 non-small cell lung cancer (NSCLC) cell lines was designed according to the arrangement specified in these data sets. However, the functions for screen evaluation actually expect this information as files on the local harddisk such as "mpi384_measure.txt", "mpi384_control.txt" and "mpi384_dilution.txt" which are installed together with the package. Default examples of these configuration files are included as "default96_measure.txt", "default384_measure.txt" etc.

In the "measure" file, there must be one row specified for each measurement series. This consists of the respective compound name followed by tab-delimited information on the wells where the measurements are located. Each of these must be given as a comma-delimited pair of coordinates. The same is expected for the control wells in the "control" file. Here, a particular well can be specified multiple times and will then be repeatedly used as a control well for the signal intensity without a compound being applied. If the normalize argument to the functions is specified as "single", there must be one control well for each single measurement; if on the other hand, "mean" is selected, an arbitrary number of wells can be specified and the mean of those values is used for normalization. Finally, the number of rows and the row names in the "dilution" file must equal those in the preceding two configuration files. Each row contains the compound name followed by a tab-delimited list of the concentrations used for the respective measurement series. It should by obvious that the number of concentrations in one row must equal the number of wells in the "measure" file for each row. However, the number of control wells can be distinct from these if the normalize argument is set to "mean" such that the mean of the respective control row is taken.

Importantly, the number of rows must be equal in all three files as well as the row names, where case-sensitivity and literal equality has to be carefully verified. The easiest way to create the configuration files is to simply start the GUI using ic50() which automatically creates a default version to be modified by the user. After having saved this configuration, it can be repeatedly used for screen evaluations as long as the experimental setup is not changed.

A step-by-step tutorial document describing how to prepare the data and configuration is included in the ic50 package.

References

Frommolt P, Thomas RK (2008): Standardized high-throughput evaluation of cell-based compound screens. BMC Bioinformatics, 9(1): 475

Sos ML, Michel K, Zander T, Frommolt P, Weiss J, et al. (2009): Predicting drug susceptibility in non-small cell lung cancers based on genetic lesions. J Clin Invest, 119(6): 1727-40