18.3. Experiments

18.3.1. Experiment properties
18.3.2. Experimental factors
18.3.3. Experiment overview
18.3.4. Tab2Mage export

Experiments are the starting point for analysis. When you have uploaded and imported your raw data, collected and registered all information and annotations about samples, hybridizations, and other items, it is time to collect everything in an experiment.

To create a new experiment you can either mark one ore more raw biossays on the raw bioassays list view and use the New experiment button. You can also create a new experiment from the experiments list view.

18.3.1. Experiment properties

Figure 18.3. Create experiment

Create experiment

Name

The name of the experiment.

Raw data type

The raw data type to use in the experiment. All raw bioassays must have raw data with this type.

Directory

A directory in the BASE file system where plug-ins can save files that are generated during the analysis. This is optional and if not given the plug-ins must ask about a directory each time they need it. Use the Select button to browse the file system or create a new directory.

Raw bioassays

The raw bioassays you want to analyze in this experiment. If you created the experiment from the raw bioassays list the selected raw bioassays are already filled in. Use the Add raw bioassays button to add more raw bioassays or the Remove button to remove the selected raw bioassays from the list.

Description

A description of the experiment.

Click on the Save button to save the changes or on Cancel to abort.

The publication tab

On this tab you can enter information about a publication that is the result of the experiment. All of this information is optional.

PubMedId

The ID of the publication in the PubMed database.

Title

The title of the publication.

Publication date

The date the article was published. Use the Calendar button to select a date from a pop-up window.

Abstract

The article abstract.

Experiment design

An explanation of the experiment design.

Experiment type

A description of the experiment type.

Affiliations

Partners and other related organisations that have helped with the experiment.

Authors

The list of authors of the publication.

Publication

The body text of the publication.

Click on the Save button to save the changes or on Cancel to abort.

18.3.2. Experimental factors

The experimental factors of an experiment are the variables you are studying in the experiment. Typically the value of an experimental factor is varied between samples or group of samples. Different treatment methods is an example of an experimental factor.

In the BASE world an experimental factor is the same as an annotation type. Since you probably have lots of annotations on your items that are not relevant for the experiment you must select the annotations types that should make up the experimental factors of the experiment.

Use the Add annotation types button to select the annotation types that should be used as experimental factors. The Remove button removes the selected annotation types.

Click on the Save button to save the changes or on Cancel to abort.

To be able to use the values of the experimental factors in the analysis of your data the values must be accessible from the raw bioassays. Since most of your annotations are probably made at the sample or biosource level the raw bioassays must inherit those annotations. Read Section 11.4.1, “Inheriting annotations from other items” for more information about this.

[Tip] Tip

Use the Experiment overview function to verify that all your raw bioassays has been annotated or inherited values for all experimental factors. If not, you should do that before starting with the analysis.

18.3.3. Experiment overview

With the Experiment overview function you can get an overview of all hybridizations, extracts, samples, annotations, etc. used in an experiment. The overview includes a lot of checks that validates your experimental setup for missing or possibly incorrect information.

You can access the overview of an experiment by navigating to the single-item view of the experiment you are interested in. Then, switch to the Overview tab that is present on that page. Here is an example of what is displayed:

Figure 18.4. The experiment overview

The experiment overview

The page is divided into three sections:

  • To the left is a tree displaying all items in the experiment and how they are linked to each other.

  • The lower right shows a list of warnings and error messages that was found when validating the experiment. In the example you can see that we have failed to specify a value for the Temperature protocol parameter for one of the samples.

  • The upper right shows information about the currently selected item in the tree. This part also contains more information errors or warnings for this item. It may also present you with one or more suggestions about how to fix the problem and with a link that takes you to the most probable location where you can fix the error or warning.

    [Note] No links?
    If you do not have permission to change things no links will be shown.

Validation options

Click on the Validation options button in the toolbar to open the Validation options dialog.

Figure 18.5. Validation options

Validation options

The validation procedure is highly configurable and you can select what you want to ignore, or something should be displayed as an error or warning.

Presets

The list contains predefined and user defined validation options. Use the Save as… button to save the current options as a user defined preset. The Remove… button is used to remove the currently selected preset. Predefined presets cannot be deleted.

Project defaults

The options in this section are used to check if your experiment uses the same values as set by the project default values of the currently active project (see Section 7.2, “Projects”). If no project is active or if the active project does not have default values these options are ignored.

Missing items

The options in this section are used to check if you have specified values for optional items. For example, there is an option that warns you if you have not specified a protocol.

Annotations

The options in this section are used to check problems related to annotations. The most important ones are listed here:

  • Missing MIAME annotation value: Checks that you have specified values for all annotations marked as Required for MIAME.
  • Missing factor value: Checks that you have specified values for all annotations used as experimental factors in the experiment.
  • Missing parameter value: Checks that you have specified values for all protocol parameters.
  • Annotation is protocol parameter: Checks if an item has been annotated with a an annotation that is actually a protocol parameter.
  • Annotation has invalid value: Checks if annotation values are correct with respect to the rules given by the annotation type. This might include numeric values that are outside the valid range, or values not in the list of allows values for an enumerated annotation type.
  • Inheriting annotation from non-parent: Checks if inherited annotations really comes from a parent item. This might happen if you rearrange parent-child relationship because you found that they were incorrectly linked.
Denied access

The options in this section are used to check if you do not have access (read permission) to an item in the experiment hierarchy. If this happens the validation cannot proceed in that branch. This might mask other validation problems.

Other

This section collects options that does not fit into any of the other sections. The most important options are:

  • Array deign mismatch: Checks if the array design specified for a raw bioassay is the same array design specified for the hybridization.
  • Multiple array designs: Checks if all raw bioassays use the same array design or not.
  • Incorrect number of labled extracts: Checks if the number of labeled extracts match the number of channels for the experiment.
  • Non-unique name: Checks if two items of the same type have the same name. A unique name if important when exporting data in Tab2Mage format.
  • Circular reference to pooled item: If you have used pooling, checks that no circular references have been created.

Click on the Save button to use the current settings. The display will automatically refresh itself.

Fixing validation failures

The experiment overview includes a function that allows you to quickly fix most of the problems found during the validation. The easiest way to use the function is:

  1. Click on an error or warning in the list of failures in the lower right pane. The tree in the left pane and the item overview in the top right pane will automatically be updated to show the exact location of the faulty item.
  2. The upper right pane should contain a list labeled Failure details with more information about each failure and also one or more suggestions for fixing the problem. For example, a failure due to a missing item should suggest that you add or select an item.
  3. The suggestions should also have links that takes you to an edit view where you can do the changes.
  4. After saving the changes you must click on the Revalidate button to update interface. If you want, you can fix more than one failure before clicking on the button.

18.3.4. Tab2Mage export

Tab2Mage format is a tab-delimited format veted by EBI's ArrayExpress repository for submission microarray data. Tab2Mage format has been chosen by BASE to provide an easy way for data deposition to public repository when submitting a manuscript and publishing experimental data.

BASE has been engineered to closely map the MIAME concepts and a number of BASE entities can be mapped directly to Tab2Mage elements. However, since MIAME is focused on microarray processing workflow, information about the biological sample is down to the user. To accommodate the annotation needs of users, BASE provides a mechanism that allows annotation customization to meet user specific requirements. BASE allows to create annotation type for quantitative annotation and qualitative annotation

BASE can export an experiment to Tab2Mage format thanks to a dedicated export plug-in. For the plug-in to work, it is important to understand that information recorded in BASE should be formatted following a small number of rules. Failing to do so may impair the possibility of exporting to ArrayExpress.

[Note] Note

The Tab2Mage export plug-in has not yet been included in the main distribution. Hopefully, it will appear in the next (2.4) release.

Biomaterial annotations

Tab2Mage specifications only allow BioSource items to be annotated with BioMaterialCharacteristics.

[Warning] Warning

All BASE Annotation Types used to annotate at the level of Sample and Extract items will be lost during the export in Tab2Mage format in order to comply with the ArrayExpress Tab2Mage parser.

[Note] Note

In the context of data exchange between BASE instances, the export function can be altered to allow attachment of annotations to items other than biosources, therefore avoiding loss of information.

Annotation units

To associate units to BASE annotation types and remain compatible with Tab2Mage specifications, users need to adhere to the following convention:

annotation_type_name(unit_name) as in body mass(kg) or concentration(mg/ml)

[Warning] Warning

Only one unit can be specified at any one time for any given annotation type. In order to enable Tab2Mage support, it might be necessary to declare several related Annotation Type in order to report similar kind of information but expressed in a different unit. Specifying Age for instance is a good example on how to deal with such cases: One should create several related annotation types e.g. Age(week), Age(year) or Age(month) as those variations maybe be necessary when reporting the age of a mouse or the age of a human volunteer.

Protocol parameters

In order to ensure MIAME compliance, Tab2Mage specifications cater for reporting parameters attached to protocols and all parameters attached to a protocol should be declared in the protocol section of a Tab2Mage file.

In BASE terms, Tab2Mage elements such as BioMaterialCharacteristics, Parameter or FactorValues are all annotation types. But, it is necessary to flag those annotations types meant to be used as protocol parameters as such so that they can identified by the Tab2Mage exporter and handled appropriately.

[Warning] Warning

It is not possible to use the same annotation type Temperature for reporting a patient body temperature (which is a Biomaterial Characteristic) and hybridization temperature (which is a protocol parameter). Hence it will be necessary to declare 2 distinct annotation types:

  • Annotation type to be used as BioSource characteristics: body temperature (degree_C)

  • Annotation type to be used as protocol parameter: hybridization temperature (degree_C)

Experimental factors

It is a MIAME requirement to identify Experimental Variables when submitting data to ArrayExpress (provided the study is an intervention study). Therefore, BASE users willing to use the Tab2Mage export function will have to declare Experimental Factors using the the Experimental Factor tab available when editing experiments. See Section 18.3.2, “Experimental factors” for more information.

Values for the experimental factors are take from annotations. The annotation must exist at the raw bioassay level, which probably means that you have to inherit the annotation from some other item, for example, a biosource or a sample. It is also possible to use a protocol parameter as experimental factor. See Chapter 11, Annotations for more information about annotations.