printlogo
http://www.ethz.ch/index_EN
Institute of Microbiology
 
print
  

O - Adjust Output

>> Output >> Adjust output

The "Generate Excelfiles" window allows the generation of “human readable” data, in the form of an Excel-file, a heat map or an outlined image. This Adjust-Output module allows defining the kind of output that is generated.

Note: in a Linux or a Macintosh environment the Matlab command "xlswrite" does not work. We therefore generate comma-separated value files which can be imported into Excel. While importing chose "comma" as the deliminator. Three files are generated: one for "image-variables", one for "well-variables" and one for "plate-variables".

Background concepts and general strategy

Please also check the "Basic concepts" earlier in the manual.

Vectors can simply be viewed as a line of binary numbers (either 0 or 1). The list is as long as the number of objects they refer to in one image. The number is 1 if on particular object is concerned and 0 if it is not. For instance, nucleus number 50 might have been classified as mitotic; therefore the number in that position would be 1. Vectors are calculated for image-wise and are saved and accessible for well- and plate variables.

Measurements: are CellProfiler measurements of a CellProfiler object; examples are measurements for object intensity, texture or shape. Measurments of interest can be saved, can be converted to a vector or used for further calculations. Measurements are also combined for Well- and Plate variables.

Image-Variables are scalar numbers, (for instance -1 or 3.9 or 100). For each image variable, one value is calculated for each image. Image variables are calculated either from vectors, from CellProfiler-objects or other image variables. The image variables are saved and are available later for calculation of Well-Variables or for export to Excel.

Well-Variables are also scalar numbers. They are calculated from Image-Variables or other Well-Variables. They are useful to summarize all images in one well. For that, Enhanced CellClassifier checks the filenames of the analyzed images. For basically all automated microscopes the position within the well-plate is a part of the filename. You can chose in the settings file the string proceeding and following the well name, for instance “_A01_”.

Important note: if well position is not contained in the filenames or is coded differently, this algorithm does not work. In that case, you could either adjust the microscope settings or ask (us) for technical support.

Plate-Variables: are scalar values, one value per plate. In order to calculate a plate variable you chose the wells you are interested, and the well variable which should be combined. Plate variables can be used to normalize well variables or can be calculated as control values.

Outlined images or Control images: When a whole plate is summarized, "outlined images" can be generated. These images will have the same settings concerning channels, intensity scaling and so on as the current setting for display in the "Main Window". In addition, the outlines of the objects are visualized by coloring. The color thereby encodes information; it can visualize vectors or the classification process. In an example, the outlines of the objects of the class "mitotic" would be red; outlines of the objects of the class "non-mitotic" blue. This allows for an easy visual control of the image analysis process in a whole set of images.

Heat maps refer to graphical summaries of Well-Variables for the whole plate. They are intended to give a broad overview over the whole plate to detect major problems at one glance.

Matlab Summary: Image-Variables and Well-Variables can be exported as a Matlab-variable. This variable has the name Sum and to fields: Image and Well. Matlab summaries are convenient for experiments with many plates (for instance a screen) which are analyzed using customized scripts.

CellClassifier output files, with the ending .yac, readable by the open source program HCDC-KNIME, and in the future possibly others are produced together with the Matlab Summary and stored in the same folder.

Excel-File: For each plate an Excel-File is generated. The file contains to sheets: Image and Well. All Image-Variables and all Well-Variables are saved to this sheet. The description you give for these variables (see below) will be the headline for the respective column. If you do not want to export a particular variable, type ‘none’ as its description.

Adjustment-Radio buttons

Adjustment radio buttons are located in the top of the Graphical User Interface. Six different outputs can be adjusted: Vectors, Measurements, Outlines, Image Variables, Well Variables, Plate Variables, Heat maps and Matlab-Summaries. Simply press the respective radio button to switch between the different outputs.

List box

For each item (Vectors, Outlines, Image-Variables etc.) many items can be generated, for instance 10 vectors, 5 image variables. For each output this number is unlimited. The list box shows the names of all items of the respective output (for instance the names of the 10 vectors if "Vector" is selected by the respective radio button, see above). The items are shown in the order they are calculated, the item calculated first above. Simply click on the desired item to make changes.

Plus-Button

Initially the list box will be empty (if no items have been defined in the settings-file). To generate a new item, press the Plus-Button (+), after this a new item is generated and filled with the default values.

Note: If a certain field for certain variables requires an input (for instance, the definition of a Well-Variable might require image variables), but this input has not been defined yet, the word ‘empty’ will replace the non-existing variable. You cannot return with these invalid settings to CC (CC will complain).

Minus-Button

Press the minus (-) button to remove the current line from the menu. You will be asked for confirmation. After deleting the last line, “empty” will appear in all fields.

Note: there is no UNDO-option for deletion of items.

Up and Down Button

To change the order of the lines, use the Up and Down buttons.

Note: CC makes plausibility checks, for instance, if an image variable is used before it is defined, a warning will be produced after each user input until the problem is corrected (using the Up and Down Pushbuttons).

The items of the AdjustOutput window: Command, Specification, Name, Description, Parameter 1 and Parameter 2

Generally, each item contains one command and at least one parameter. Some vector commands in addition require a command specification; some commands require 2 parameters (parameter 1 and parameter 2). Furthermore, depending on the kind of output an item has a name and in some cases a field for a description. All these adjustments are done with the graphical user interface. The interface is dynamic; impossible entries are almost completely prevented.

For each command, the usage is explained below. In addition, for all commands a short explanation is displayed in the lower part of the window.

Note: While you have freedom in choosing basically any input for a description field, a field which is later used as a Matlab-variable accepts only a limited input, CC will complain about invalid entries. Especially space characters and special characters (*%&…) are not allowed.

Measurements

Name of the Measurements: should be a variable name (no spaces, no special characters, see above).

Command: Currently only the command 'Copy' is implemented.

Parameter 1: is the name of the CellProfiler object which has been measured

Parameter 2: is the measurement of this CellProfiler object you are interested in

Parameter 3: is a feature of this measurement

Only valid choices for the 3 parameters are offered in the popup-menus.

Vectors

Name of the Vector should be a variable name (no spaces, no special characters, see above).

Original CP-object is the CellProfiler object from which the vector will derive. This determines the size of the vector in a given image. In addition, some commands use measurements from this object for further calculations (for instance, for the command “Children” these are the parent objects). A number of choices are given in form of a pop-up-menu

Original vector: Choose either "all" or the name of another (=original) vector. After choosing "all", all calculations start with a vector containing only ones (for instance 1; 1; 1). Choosing another vector as the original vector enables exclusion of objects from the resulting vector. For instance, if the original vector had been 0; 1; 0, the resulting vector will always have 0 in positions 1 and 3.

Warning: If another vector is chosen, this other vector must have the same size as the current vector. This will always be the case if the both vectors have the same CellProfiler-object as its parent. Alternatively, if both vectors derive from CellProfiler objects which have a parent-child relationship (Nuclei, Cells and Cytoplasm in most CellProfiler setups) combinations are also possible. Two vectors derived from completely independent objects cannot be combined.

Command and Command-Specification: The choice of the currently implemented commands is given in the form of a pop-up menu

Parameter 1 and Parameter 2 are different for each command.

Calculations always start with the original Vector, which is modified.

Command ‘Not’: Parameter 1 is a vector, this vector is now subtracted from the original vector. Thereby, positions where this vector had been ‘true’ = 1 are excluded from the original vector (= are zero in the resulting vector).

Command ‘Class’: Parameter 1 is the name of the class. All positions where the original vector was ‘true’ and this class was present will be ‘true’ in the resulting vector.

Command ‘Neighbor’: The objects of column 2 should have been measured with the MeasureNeighbor CP-module.

Command ‘Children’: The objects of Parameter 1 (children) should be the "Children" of the Original CP-object ("parent" object). "Parent" - "Children" relationships exist between primary and secondary or tertiary objects or can be generated with the "Relate" CP-module. The popup-menu for Parameter 1 offers all valid children objects available for the chosen original CP-object.

Note: Please take care not to confuse vectors from different CP-objects. If as a result of such confusion the vectors have different length, an error while calculating the Excel-file will ensue. We detailed above (when discussing the original vector), which vectors can be savely combined.

Command "Threshold":
Parameter 1 is a Cell-Profiler measurement which must have been selected via "Measurements" in this window. Parameter 2 is a threshold, either a lower threshold or a higher (upper) threshold.

Outlines

In this window, the outlines of the control images can be defined.

Outlines can either be generated directly with the information from the classifier, using the "Class" command or with a vector (command "Vector"). In theory, overlapping vectors can be used; an overlap between a vector and a class is also tolerated. However, the command applied last will paint over the labels of already colored objects.

Command Class: Parameter 1 is the name of the class.

Command Vectors: Outlines are generated in the respective color if the vector is "true" for the respective object and ignored if the vector is "false".

The current color is indicated with the color button. Please choose the color, either via pressing the button, or via typing in the color in the RGB-format (3 values for red, green and blue between 0 and 1) or by typing in the name of the color. Accepted inputs are: ‘yellow’, ‘magenta’, ‘cyan’, ‘red’, ‘green’, ‘blue’, ‘white’, ‘black’ or: ‘y’, ‘m’, ‘c’, ‘r’, ‘g’, ‘b’, ‘w’, ‘k’. Other inputs will be ignored.

Image Variables

With the help of this window, image based numbers can be generated (this means, the result is one number per image). These numbers can be displayed in the resulting Excel-sheet (if desired). For calculation of these Image-variables, CellProfiler data, classification data, other image variables and vectors can be used.

The first field is a description.

Note: if you type in "none", this variable will not be saved to the Excel-sheet

The second field is a variable name; only limited inputs will be accepted. The next fields are "Command", "Parameter 1" and "Parameter 2".

Command "ObjectCount": The Parameter 1 is a CellProfiler object (for instance nuclei or cells). The Image Variable is the sum of these objects in the respective image.

Command "Vectors": The Parameter 1 is a vector. CC simply counts how many objects within the vectors are "true". The resulting value is the sum of "true" positions.

Command "Percent": Parameter 1 and Parameter 2 are image variables. The variable specified by Parameter 2 corresponds to 100%. The result is calculated: Parameter 1/ Parameter 2 * 100.

Command "Calculate": Parameter 1 is a measurement; parameter 2 is either "min", "max", "median" or "stdev". The resulting image variable will be the minimum, maximum, median or standard-deviation of the selected measurements for this image.

Command "Granularity": Parameter 1 is the name of the CP-image which has been measured by the granularity method, (for instance "OrigBlueSpectrum" or "RescaledBlueSpectrum", a choice is given with a pop-up-menu). The second parameter is either "max" or "first_value" (see explanations to the Focus-window).

Command "Focus": no Parameters. The resulting value will be 1 if the image has been in focus, 0 for an out of focus image and -1 if the Focus-mode is "off".

Command "Threshold": The threshold will be extracted. Parameter 1 is the CP-object for which the threshold is of interest. Parameter 2 is not used.

Well Variables

With the help of this window, well based numbers can be generated (this means, the result is one number per well). These numbers can be displayed in the resulting Excel-sheet (if desired). For calculation of these well-variables, image variables and other well variables can be used.

Note: While generating well based variables, CC loops over rows A-P and columns 1-24. For probably all automated microscopes the combination of row and column and underscores (for instance [String1]A01[String2]’) are part of the filename, this is how the images belonging to the same well are recognized. Usually String1 and String2 are underscore '_'. This can be adjusted as described for "Identification of Wells" in the "DefineSettings" window. If this does not work for your microscope filenames, contact (us) for technical support.

The first field is a description.

Note: if you type in "none", this variable will not be saved to the Excel-sheet

The second field is a variable name; only limited inputs will be accepted. The next fields are for "Command", "Parameter 1" and "Parameter 2".
Currently implemented commands: "AddVector", "AddImageVariable", "Percent", "Calculate", "log2", "log10", "Normalize".

Command "AddVector": Parameter 1 is the name of the Vector which should be summarized. Parameter 2 is not used. The command simply summarizes the vector for this well (how often the vector had been "true" =1).

Command "AddImageVariable": Parameter 1 is the name of the Image Variable which should be summarized. Parameter 2 is not used. The command simply generates the sum of all image variables for the images which belong to this well.

Command "Percent’: Parameter 1 and Parameter 2 are predefined well variables. Percent is calculated: Parameter_1/ Parameter_2 * 100.

Command "Calculate: Parameter 1 is a measurement, parameter 2 is either "min", "max", "median" or "stdev". The resulting well variable will be the minimum, maximum, median or standard-deviation of the selected measurements for this well.

Command "Normalize": Parameter 1 is a well variable, parameter 2 is a plate variable. The resulting number will be: well variable / plate variable. Well variables can thereby be expressed for instance relative to several control wells.

Commands "Log2" and "Log10": The logarithm of the well variable to the basis 2 or 10 is calculated

Plate Variables

With the help of this window, plate based numbers can be generated (this means, the result is one number per plate). In general, a plate variable will summarize a well variable for a choice of wells (for instance, Mock-treated cells from wells "A01", "A02" and "A03"). Plate variables are exported to the resulting Excel-sheet and can be used to normalize Well-variables.

Description: any input is accepted and will be the headline in the Excel-sheet

Name of plate variable: internal reference within CC (no spaces, no special characters)

Well variable: the well variable which should be summarized.

Wells: One can either type in the well-names (letters followed by 2-digit numbers separated by commas (for instance: 'A01, A02, A03') or select the wells a graphical user interface after pressing the "select" button.

Command "Add": The resulting plate variable will be the sum of the well variables for the selection of wells.

Command "Median": The resulting plate variable will be the median of the well variables for the selection of wells.

Heat maps

With the help of this window, the design of the heat map-sheet(s) is defined. Heat maps are meant to give a quick overview over a whole plate. Other kinds of diagrams are also possible.

Description: Will be displayed as a headline above the diagrams. Choose a short descriptive name (for instance "Number Nuclei"). The other fields define the Command, Parameter 1 and Parameter 2 or a Modification.

Command Heatmap: a heat map is a 2-dimensional display of the plate, each well is a little field, and the color of the respective field gives the values, specified by a colorbar. This display is ideal for a quick overview over the results of an experiment. Parameter 1 gives the name of the well-variable which should be displayed as a heat map. Parameter 2 defines the modification of the well variable and is either

Command Histogram: Parameter 1 describes Well-Variable 1. Parameter 2 is not used. A histogram for the well variable is calculated and displayed; Parameter 1 will also be the label for the y-axis.

Comman Lowess: Parameter 1 is one Well-Variable, Parameter 2 another. Now Well-Variable 1 (x-scale) is plotted against Well-Variable 2 (y-scale). A "Lowess" normalization is calculated and plotted (red dots) and shows local trends for the data. The Lowess-command uses a sliding window to predict trends and only works if the full-Matlab version is installed.

Matlab Summary

With the help of this window, data can be saved for later processing in a Matlab-readable format. The command lets you decide between image and well variables, Parameter 1 allows deciding the specific variable. You can save as many variables as you want.

Command Image_Variable

Command Well_Variable

Note: Enhanced CellClassifier output files (csv files with the ending .yac) are produced together with the Matlab summary, in the same folder. These files with the ending .yac can be imported by the program HCDC-KNIME, and in the future possibly others. 

 

Wichtiger Hinweis:
Diese Website wird in älteren Versionen von Netscape ohne graphische Elemente dargestellt. Die Funktionalität der Website ist aber trotzdem gewährleistet. Wenn Sie diese Website regelmässig benutzen, empfehlen wir Ihnen, auf Ihrem Computer einen aktuellen Browser zu installieren. Weitere Informationen finden Sie auf
folgender Seite.

Important Note:
The content in this site is accessible to any browser or Internet device, however, some graphics will display correctly only in the newer versions of Netscape. To get the most out of our site we suggest you upgrade to a newer browser.
More information

© 2014 Microbiology ETH Zürich | Imprint | Disclaimer | 23 December 2009
top