When data is uploaded from your computer or analysis is done on existing data using Galaxy, each output from those steps generates a dataset. These datasets (and the output datasets from subsequent analysis on them) are stored by Galaxy in Histories.
All users have one 'current' history, which can be thought of as a workspace or a current working directory in bioinformatics terms. You current history is displayed in the right hand side of the main 'Analyze Data' Galaxy interface in what is called the history panel.
|Figure 1. Galaxy history is simply the right panel of the interface|
The history panel displays output datasets in the order in which they were created with the oldest/first shown at the bottom. As new analyses are done and new output datasets are generated, the newest datasets are added to the top of the history panel. In this way, the history panel displays the history of your analysis over time.
Users that have registered an account and logged in can have many histories and the history panel allows switching between them and creating new ones. This can be useful to organize different analyses.
Anonymous users (if your Galaxy allows them) are users that have not registered an account. Anonymous users are only allowed one history. On our main, public Galaxy server, users are encouraged to register and log in with the benefit that they can work on many histories and switch between them.
Above the current history panel are three buttons: refresh, history options, and 'view all histories'
|Figure 2. History controls.|
These history controls perform the following functions:
- The refresh button will force a reload of the history being viewed. This can be helpful if you believe the history interface needs to be updated or is otherwise not working properly.
- The history options button opens a secondary menu allowing you to perform other history-related tasks.
- The 'view all histories' button sends you to the interface for managing multiple histories.
Histories also store information in addition to the datasets they contain. They can be named/re-named, tagged, and annotated.
All histories begin with the name 'Unnamed history'. Non-anonymous users can rename the history as they see fit:
- Click the existing name. A text input field will appear with the current name.
- Enter a new name or edit the existing name in the way you'd like.
- Press 'Enter' to save the new name. The input field will disappear and the new name will display immediately.
- To cancel renaming, press 'ESC' or click outside the input field.
|Figure 3. Renaming a history by clicking on its name and entering a new one.|
Tags are short pieces of text used to describe the thing they're attached to and many things in Galaxy can be tagged. Each item can have many tags and you can add new tags or remove them at any time. Tags can be another useful way to organize and search your data. For instance, you might tag a history with the type of analysis you did in it: 'assembly' or 'variants'. Or you may tag them according to data sources or some other metadata: 'long-term-care-facility' or 'yellowstone park:2014'.
To tag a history:
- Click the tag button at the top of the history panel. An input field showing existing tags (if any) will appear.
- Begin typing your new tag in the field. Any tags that you've used previously will show below your partial entry - allowing you to use this 'autocomplete' data to re-use your previous tags without typing them in full.
- Press enter or select one of the previous tags with your arrow keys or mouse.
- To remove an existing tag, click the small 'X' on the tag or use the backspace key while in the input field.
|Figure 4. Tagging a history will help searching for it later on.|
Sometimes tags and names are not enough to describe the work done within a history. Galaxy allows you to create history annotations: longer text entries that allow for more formatting options. Newlines and whitespace are preserved. Later, if you publish or share the history, the annotation will be displayed automatically - allowing you to share additional notes about the analysis.
To annotate a history:
- Click the annotation button at the top of the history panel. A larger text section will appear displaying any existing annotation (or, if there's none, italic text saying you can click on the control to create an annotation).
- Click the annotation section. The larger input field will appear.
- Add any annotations you desire. 'Return' will create a line break and white space is preserved. (Tabs cannot be entered since the 'Tab' button is used to switch between controls on the page - tabs can be pasted in however).
- To save the annotation, click the 'Done' button.
|Figure 5. Annotating a history allows entering more information such as, for example, experimental details related to the analysis.|
As datasets are added to a history, Galaxy will store them as files on its file system. The total size of these files for all the datasets in a history is displayed underneath the history name. For example, if a history has 200 megabytes of data on Galaxy's filesystem, '200 MB' will be displayed underneath the name.
If your Galaxy server uses quotas, the total combined size of all your histories will be compared to your quota and if you're using more than the quota allows Galaxy will prevent you from running any new jobs until you've deleted some datasets and brought that total below the quota.
Datasets in the history panel show the state of the job that has generated or will generate the data.
There are several different 'states' a dataset can be in:
- When you first upload a file or run a tool, the dataset will be in the queued state. This indicates that the job that will create this dataset has not yet started and is in line to begin.
- When the job starts the dataset will be in the running state. The job that will create these datasets is now running on Galaxy's cluster.
- When the job has completed successfully, the datasets will be in the ok state.
- If there's been an error while running the tool, the datasets will be in the error state.
- If a previously running or queued job has been paused by Galaxy, the dataset will be in the paused state. You can re-start/resume paused jobs using the options menu above the history panel and selecting 'Resume Paused Jobs'.
|Figure 6. Dataset states.|
Datasets in the panel are initially shown in a 'summary' view that only displays:
- a number indicating in what order (or what step) this dataset was created
- the dataset name
- a view button: click this to view the dataset contents in raw format in the browser
- an edit button: click this to edit dataset properties
- a delete button: click this to delete the dataset from the history (don't worry, you can undo this action)
|Figure 7. Controls for "summary" (or collapsed) dataset view.|
You can click the dataset name and the view will expand to show more details:
- a short description of the data
- the file format
- the reference sequence (or database) for the data
- (optionally) some information/output from the job that produced this dataset
- a row of buttons that allow further actions on the dataset
- a peek of the data: a couple of rows of data with the column headers (if available)
|Figure 8. Controls for expanded dataset view.|
There are two types of tags that can be used as an additional level of labeling for datasets: standard tags and hashtags (also known as propagating tags). The standard tags work similarly to history tags described above (Fig. 4) - they add another level of description to datasets making them easier to find:
|Figure 9. Standard tags provide an additional level of annotation for individual datasets. A. Tags are added by clicking on the tags icon and entering a name. B. Here the tag is used to search the history. Entering
Hashtags are much more powerful as they are displayed in the history panel and propagate through the analysis:
|Figure 10. Hashtags allow you to more easily track datasets through the analysis. A. Hashtags are added similarly to standard tags but with one important difference: they are prepended with a hash
The following video highlights hashtags in action:
Some procedures in Galaxy such as workflows will often hide history datasets in order to simplify the history and hide intermediate steps of an automated analysis. These hidden datasets won't normally appear in the history panel but they are still mentioned in the history subtitle (the smaller, grey text that appears below the history name). If your history has hidden datasets, the number will appear there (e.g. '3 hidden') as a clickable link. If you click this link, the hidden datasets are shown. Each hidden dataset has a link in the top of the summary view that allows you to unhide it. You can click that link again (which will now be 'hide hidden') to make them not shown again.
|Figure 11. Hiding and unhiding datasets. Left side shows a history with one hidden dataset. We know this because a link "1 hidden" appears under the history's name. Clicking this link will reveal the hidden dataset as shown on the right side of the figure.|
You can delete any dataset in your history by clicking the delete button. This does not immediately remove the dataset's data from Galaxy and it is reversible. When you delete a dataset from the history, it will be removed from the panel but (like hidden datasets) the total number of deleted datasets is shown in the history subtitle as a link. Clicking this link (e.g. '3 deleted') will make the deleted datasets visible and each deleted dataset will have a link for manually undeleting it above its title. You can click that link again (which will now be 'hide deleted') to make them not shown again.
|Figure 12. Deleting and undeleting datasets. The left side shows a history with one deleted dataset. We know this because a link "1 deleted" appears under the history's name. Clicking this link will reveal the deleted dataset as shown on the right side of the figure. From here it can be undeleted or deleted permanently.|
If you are showing deleted datasets and your Galaxy allows users to purge datasets, you will see an additional link in the top of each deleted dataset titled 'Permanently remove it from disk'. Clicking this will remove the file that contains that dataset's data and will decrease the disk space used by the history. This action is not reversible and cannot be undone.
If your Galaxy doesn't allow users to purge their datasets, you will not see that link.
Depending on the policy of your Galaxy server, administrators will often run scripts that search for and purge the datasets you've marked as deleted. Often deleted datasets and histories are purged based on the age of the deletion (e.g. datasets that have been marked as deleted for 90 days or more). Check with the administrators of your instance to find out the policy used.
You can also hide, delete, and purge multiple datasets at once by multi-selecting datasets:
- Click the multi-select button containing the checkbox to the right of the history size.
- Checkboxes will appear inside each dataset in the history.
- Scroll and click the checkboxes next to the datasets you want to manage. You can also 'shift+click' to select a range of datasets.
- Click the 'All' button to select all datasets. Click the 'None' button to clear/remove all selections.
- Click the 'For all selected...' to choose an action. The action will be performed on all selected datasets. (If an action doesn't apply to a selected dataset - like deleting a deleted dataset - nothing will happen.)
- You can click the multi-select button again to hide the checkboxes again.
|Figure 12. Operating on multiple datasets can be enabled by clicking on the checkbox icon .|
You can filter which datasets are shown and search for datasets using the search bar at the top of the panel. Enter any text that a dataset you'd be looking for would contain, including:
- the name or part of the name
- any text (or partial text) from the description and info fields
- the file format or reference database
- any text or partial text from the annotation or tags of a dataset
- To find all VCF files you might enter:
- To find all files whose names contain data 1, you can enter:
- To search for a VCF file named 'VCF filter on data 1' and tagged with 'experiment 1', you could enter:
vcf filter on data 1 experiment 1
|Figure 14. Searching for datasets.|
You can clear a search and show all visible datasets by clicking the round 'X' button in the right of the search bar or - while entering text in the search bar - hitting the escape key ('ESC').
You can also specify what dataset properties you're searching using keyword search. These are the property names
followed by '=' then the value. When using these only the property named is searched for that value. Keyword searches can be combined:
database=mm10 annotation=successful and you can enclose text and include spaces using double quotes:
name="My Dataset" annotation="First run".
All keyword searches are connected with logical ANDs which means all searches must be true/hit in order for the dataset to be shown.
name=my-datasetwould show all datasets whose names contain 'my-dataset'
format=vcf: to search for datasets in a particular format
database=hg19to search for all datasets with a particular reference set
annotation="first of five"for the dataset annotation
description="This is data of a Borneo Orangutan"for dataset summary description
info="started mapping"for searching on job info
tag=experiment1for searching on (a partial) dataset tag. You can repeat to search for more tags.
hid=25to search for an item with a specific history id (which usually corresponds to its order in the history)
state=errorto show only datasets in a given state
|Figure 15. Advanced search.|
It's often useful to combine search and multi-select. Multi-selections will persist over searches and the All/None buttons will only apply to those datasets that are currently shown with the given search.
So, for example, to select and manipulate two different sets of datasets: a) fastqsanger files tagged with 'low quality' and b) hg19 reference BAM files whose names contain "Output":
- In the search bar, enter:
format=fastqsanger tag="low quality"and hit enter.
- Click the multi-select button to show the checkboxes.
- Click the 'All' button to select all the fastqsanger files
- In the search bar again, enter:
database=hg19 format=BAM name=outputand hit enter.
- Click the 'All' button again to select all the BAM files.
- Clear the search using the clear button. All datasets are shown and both fastqsanger and BAM files are still selected.
- You can now perform some action on those two sets of datasets.
If you have not purged a history and have only deleted it, it is possible to 'undelete' it and reverse or undo the deletion. Since one of the purposes of deleting histories is to remove them from view, we'll use the interface to specifically search for deleted histories and then to undelete the one we're interested in.
There are two ways to do this currently: via the multi-history panel and through the saved histories page. The multi- history method is presented here:
|Click the multi-history icon at the top right of the 'Analyze Data' (home) page. Note: you must be logged in to see the icon and use the multi-history page. You should see all the (non-deleted) histories that you've created.|
|Click the '...' icon button in the grey header at the top of the page. You should see a dialog that presents some options for viewing the histories. Click the 'include deleted histories' option.|
|The page should reload and now both non-deleted and deleted histories will be displayed. Deleted histories will have a small message under their title stating 'This history has been deleted'.|
|Now click the small button with the down arrow just above the deleted history you want to undelete. Then click the 'Undelete' option there. Your history should now be undeleted.|
|Click the 'Switch to' button at the top of that history and then click 'done' at the very top left to return to the 'Analyze Data' page.|
When you have multiple datasets that will be sent through the same analysis it can often be useful to place those datasets in a dataset collection. When collections are used as input when running a tool, you're telling Galaxy to run that tool on each dataset in the collection using the same settings. This happens automatically and there's no need to fill in the tool form more than once. The following tutorial describes collections in depth.