Galaxy 101 for everyone

Overview
Questions:
  • What are the differences between the Iris species?

Objectives:
  • Familiarize yourself with the basics of Galaxy

  • Learn how to obtain data from external sources

  • Learn how to tag datasets

  • Learn how to run tools

  • Learn how histories work

  • Learn how to create a workflow

  • Learn how to share your work

Time estimation: 1 hour 30 minutes
Level: Introductory Introductory
Supporting Materials:
Last modification: Oct 18, 2022
License: Tutorial Content is licensed under Creative Commons Attribution 4.0 International License The GTN Framework is licensed under MIT

Introduction

This practical aims at familiarizing you with the Galaxy user interface. It will teach you how to perform basic tasks such as importing data, running tools, working with histories, creating workflows and sharing your work. Not everyone has the same background and that’s ok!

Comment: Results may vary

Your results may be slightly different from the ones presented in this tutorial due to differing versions of tools, reference data, external databases, or because of stochastic processes in the algorithms.

Agenda

In this tutorial, we will cover:

  1. Introduction
    1. What does Galaxy look like?
  2. Create a history
    1. Upload the Iris dataset
  3. Pre-processing
    1. Convert format
    2. Remove header
  4. Data Analysis: What does the dataset contain?
    1. How many different species are in the dataset?
    2. How many samples by species are in the dataset?
  5. Analysis: How to differentiate the different Iris species?
    1. Generate summary and descriptive statistics with
    2. Visualize Iris dataset features with two-dimensional scatterplots
  6. Galaxy management
    1. Convert your analysis history into a workflow
    2. The workflow editor
    3. Run workflow on different data
    4. Share your work
  7. Conclusion
Comment: Background

The Iris flower data set, also known as Fisher’s or Anderson’s Iris data set, is a multivariate dataset introduced by the British statistician and biologist Ronald Fisher in his 1936 paper (Fisher 1936). Each row of the table represents an iris flower sample, describing its species and the dimensions in centimeters of its botanical parts, the sepals and petals. You can find more detailed information about this dataset on its dedicated Wikipedia page.

What does Galaxy look like?

Hands-on: Log in or register
  1. Open your favorite browser (Chrome/Chromium, Safari, or Firefox, but not Internet Explorer/Edge!)
  2. Browse to a Galaxy instance of your choice
  3. Choose Login or Register from the navigation bar at the top of the page
  4. If you have previously registered an account with this particular instance of Galaxy (user accounts are not shared between public servers!), proceed by logging in with your registered public name, or email address, and your password.

    If you need to create a new account, click on Register here instead.

Comment: Different Galaxy servers

The particular Galaxy server that you are using may look slightly different than the one shown in this training. Galaxy instance administrators can choose the exact version of Galaxy they would like to offer and can customize its look and feel to some extent. The basic functionality will be rather similar across instances, so don’t worry!

The Galaxy interface consists of three main parts:

  1. The available tools are listed on the left
  2. Your analysis history is recorded on the right
  3. The central panel will let you run analyses and view outputs

Galaxy interface screenshot showing history panel on the right, tools panel on the left, and main panel at the center.

Create a history

Galaxy allows you to create analysis histories. A history can be thought of as an electronic experimental lab book; it keeps track of all the tools and parameters you used in your analysis. From such a history, a workflow can be extracted; this workflow can be used to easily repeat the analysis on different data.

Think of a workflow as a cooking recipe with a list of ingredients (datasets) and a set of instructions (pipeline of operations) that describes how to prepare or make something (such as a plot, or a new dataset). The order of operations is important as very often the next operation takes as input the result of the previous operations. For instance, when baking a cake, you would first sift the flour and then mix it with eggs as it would be impossible to sift the flour afterward. That is what we call a pipeline. To make a full meal, we may need to combine multiple recipes (pipelines) together.

The finalized pipelines can be generalized as a workflow. If we use cooking as an analogy, a workflow could represent an entire menu with all the recipes for each meal. In other words, using a workflow makes it possible to apply the same procedure to a different dataset, just by changing the input.

Hands-on: Create history
  1. Make sure you start from an empty analysis history.

    Click the new-history icon at the top of the history panel.

    If the new-history is missing:

    1. Click on the galaxy-gear icon (History options) on the top of the history panel
    2. Select the option Create New from the menu
  2. Rename your history to be meaningful and easy to find. For instance, you can choose Galaxy 101 for everyone as the name of your new history.

    1. Click on Unnamed history (or the current name of the history) (Click to rename history) at the top of your history panel
    2. Type the new name
    3. Press Enter

Upload the Iris dataset

Hands-on: Data upload
  1. Import Tool: upload1 the file iris.csv from Zenodo or from the data library (ask your instructor)

    https://zenodo.org/record/1319069/files/iris.csv
    
    • Copy the link location
    • Open the Galaxy Upload Manager (galaxy-upload on the top-right of the tool panel)

    • Select Paste/Fetch Data
    • Paste the link into the text field

    • Press Start

    • Close the window

    As an alternative to uploading the data from a URL or your computer, the files may also have been made available from a shared data library:

    • Go into Shared data (top panel) then Data libraries
    • Navigate to the correct folder as indicated by your instructor
    • Select the desired files
    • Click on the To History button near the top and select as Datasets from the dropdown menu
    • In the pop-up window, select the history you want to import the files to (or create a new one)
    • Click on Import
  2. Rename galaxy-pencil the dataset to iris

    • Click on the galaxy-pencil pencil icon for the dataset to edit its attributes
    • In the central panel, change the Name field
    • Click the Save button
  3. Check the datatype
    • Click on the history item to expand it to get more information.
    • The datatype of the iris dataset should be csv.
    • Change galaxy-pencil the datatype if it is different than csv.
      • Option 1: Datatypes can be autodetected
      • Option 2: Datatypes can be manually set
    • Click on the galaxy-pencil pencil icon for the dataset to edit its attributes
    • In the central panel, click on the galaxy-chart-select-data Datatypes tab on the top
    • Click the Auto-detect button to have Galaxy try to autodetect it.
    • Click on the galaxy-pencil pencil icon for the dataset to edit its attributes
    • In the central panel, click on the galaxy-chart-select-data Datatypes tab on the top
    • Select csv
      • tip: you can start typing the datatype into the field to filter the dropdown menu
    • Click the Save button
  4. Add an #iris tag galaxy-tags to the dataset

    • Click on the dataset
    • Click on galaxy-tags Edit dataset tags
    • Add a tag starting with #

      Tags starting with # will be automatically propagated to the outputs of tools using this dataset.

    • Check that the tag is appearing below the dataset name

    Make sure the tag starts with a hash symbol (#), which will make the tag stick not only to this dataset, but also to any results derived from it. This will help you make sense of your history.

Pre-processing

Often, one or more data pre-processing step(s) may be required to proceed with the analysis. In our case, the tools we will use require tab-separated input data and assume there is no header line. Since our data is comma-separated and has a header line, we will have to perform the following pre-processing steps to prepare it for the actual analysis:

  • Format conversion
  • Header removal

Convert format

First, we will convert the file from comma-separated to tab-separated format. Galaxy has built-in format converters we can use for this.

Hands-on: Converting dataset format
  1. Convert galaxy-pencil the CSV file (comma-separated values) to tabular format (tsv; tab-separated values)

    • Click on the galaxy-pencil pencil icon for the dataset to edit its attributes
    • In the central panel, click on the galaxy-gear Convert tab on the top
    • Select Convert CSV to Tabular
    • Click the Create dataset button to start the conversion.
  2. Rename galaxy-pencil the resulting dataset to iris tabular

    • Click on the galaxy-pencil pencil icon for the dataset to edit its attributes
    • In the central panel, change the Name field
    • Click the Save button
  3. View the generated file by clicking on the galaxy-eye (eye) icon

    Question

    How many header lines does our file have?

    The file has one header line, it contains the column names.

Remove header

Now it is time to run your first tool! We saw in the previous step that our file has 1 header line. This line does not contain any data, but the names of each column. We will now remove that line from our file before moving on to our analysis.

Comment: Tip: Finding your tool

Different Galaxy servers may have tools available under different sections, therefore it is often useful to use the search bar at the top of the tool panel to find your tool.

Additionally different servers may have multiple, similarly named tools which accomplish similar functions. When following tutorials, you should use precisely the tools that they describe. For real analyses, however, you will need to search among the various options to find the one that works for you.

Hands-on: Removing header
  1. Remove Beginning Tool: Remove beginning1 with the following parameters:
    • Remove first: 1 (to remove the first line only)
    • param-file “from”: select the iris tabular file from your history
    • Click Execute
    Comment: Tip: search for the tool

    Use the tools search box at the top of the tool panel to find Remove beginning tool.

    Settings for the `Remove beginning` tool.

  2. Rename galaxy-pencil the dataset to iris clean

    • Click on the galaxy-pencil pencil icon for the dataset to edit its attributes
    • In the central panel, change the Name field
    • Click the Save button
  3. Click on the new history item to expand it

    Question
    1. Which tags are present on this resulting dataset? (You may have to refresh the history panel to see the tags)
    2. How many samples (lines) does our dataset contain?
    1. The output of Remove beginning tool is also tagged with the label iris. Tags beginning with a hashtag (#) will propagate; they will appear on any datasets derived from your original tagged file.

    2. There are 150 lines in our file (we can see this under the file name when we have expanded the history item). This means we have 150 samples.

  4. View galaxy-eye the contents of the resulting file.
    • You should see that the header line is now no longer present.

Data Analysis: What does the dataset contain?

Now we are going to inspect the dataset using simple tools in order to get used to the Galaxy interface and answer basic questions.

How many different species are in the dataset?

In order to answer this question, we will have to look at column 5 of our file, and count how many different values (species) appear there. There are several ways we could do this in Galaxy. One approach might be to first extract this column from the file, and then count how many unique lines the file contains. Let’s do it!

Hands-on: Extract species
  1. Cut Tool: Cut1 columns from a table with the following parameters:
    • “Cut columns”: c5
    • “Delimited by”: Tab
    • param-file “From”: iris clean dataset
  2. Rename galaxy-pencil the dataset to iris species column

    • Click on the galaxy-pencil pencil icon for the dataset to edit its attributes
    • In the central panel, change the Name field
    • Click the Save button
  3. View galaxy-eye the resulting file

  4. Unique Tool: toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_sorted_uniq/1.1.0 occurrences of each record with the following parameters:
    • param-file “File to scan for unique values”: iris species column (the output from Cut tool)
  5. Rename galaxy-pencil the dataset to iris species

  6. View galaxy-eye the resulting file

    Question
    1. How many different species are in the dataset?
    2. What are the different Iris species?
    1. There are 3 species.
    2. The 3 different Iris species are:
      • setosa
      • versicolor
      • virginica

Now we have our answer! There are 3 different Iris species in our file.

Like we mentioned before, there are often multiple ways to reach your answer in Galaxy. For example, we could have done this with just a single tool, Group tool as well. (This tool also can be searched for by the term “grouping” )

Hands-on: Exercise: Grouping dataset
  1. Try answering this question (how many Iris species are in the file?) again, using a different approach:
    • Tool: Group data by a column and perform aggregate operation on other columns tool
    • Input dataset: iris clean dataset to answer the same question.
  2. Did you get the same answer as before?

  3. Rename galaxy-pencil the dataset to iris species group
  1. Group Tool: Grouping1 with the following parameters:
    • “Select data” select iris clean dataset
    • “Group by column”: Column: 5
  2. This approach should give the same answer. There are often multiple ways to do a task in Galaxy, which way you choose is up to you!

How many samples by species are in the dataset?

Now that we know that there are 3 different species in our dataset, our next objective is determining how many samples of each species we have. To answer this, we need to look at column 5 again, but instead of just determining how many unique values there are, we need to count how many times each of them occurs.

You may have noticed there were more parameters in the Group tool tool that we did not use. Let’s have a closer look and see if any of them might help us answer this question.

Comment: Tool Help

To find out more about how a tool works, look at the help text (below the Execute button).

Look at the tool help for the Group tool. Do you see any parameters that could help answer this question?

Looking at the tool help for Group tool, we see that we can also perform aggregate operations such as mean, median, sum, max, min, count (and more). Counting sounds just like what we need, let’s try it!

Hands-on: Grouping dataset and adding information
  1. Re-run galaxy-refresh the Group tool with the following parameters:
    • param-file “Select data”: iris clean
    • param-select “Group by column”: Column: 5
    • param-repeat “Insert operation”
      • “Type”: Count
      • “On column”: Column: 1
    1. Expand one of the output datasets of the tool (by clicking on it)
    2. Click re-run galaxy-refresh the tool

    This is useful if you want to run the tool again but with slightly different paramters, or if you just want to check which parameter setting you used.

  2. Rename galaxy-pencil the dataset to iris samples per species group

  3. View galaxy-eye the resulting file.

    Question

    How many samples per species are in the dataset?

    We have 50 samples per species:

    1 2
    setosa 50
    versicolor 50
    virginica 50

Analysis: How to differentiate the different Iris species?

Our objective is to find what distinguishes the different Iris species (Figure 1). We know that we have 3 species of iris flowers, with 50 samples for each:

  • setosa
  • versicolor
  • virginica

These species look very much alike as shown on the figure below.

Three species of Iris flowers.
Figure 1: Three species of Iris flowers (Image attributions: versicolor by Danielle Langlois licensed under CC BY-SA 3.0, retrieved from WikiMedia; virginica by Christer Johansson licensed under CC BY-SA 3.0, retrieved from WikiMedia; setosa by and used with permission of Sonja Keohane, retrieved from www.twofrog.com)

And our objective is to find out whether the features we have been given for each species can help us to highlight the differences between the 3 species.

In our dataset, we have the following features measured for each sample:

  • Petal length
  • Petal width
  • Sepal length
  • Sepal width
Comment: petal and sepal

The image below shows you what the terms sepal and petal mean.

Iris flower image showing Petal and Sepal.
Figure 2: Sepal and petal of Iris flowers

Generate summary and descriptive statistics with

Hands-on: Get the mean and sample standard deviation of Iris flower features
  1. Datamash Tool: toolshed.g2.bx.psu.edu/repos/iuc/datamash_ops/datamash_ops/1.1.0 with the following parameters:
    • param-file “Input tabular dataset”: iris tabular
    • “Group by fields”: 5
    • “Input file has a header line”: Yes
    • “Print header line”: Yes
    • “Sort input”: Yes
    • “Print all fields from input file”: No
    • “Ignore case when grouping”: Yes
    • In “Operation to perform on each group”:
      • param-repeat “Insert Operation to perform on each group”
        • “Type”: Mean
        • “On column”: c1
      • param-repeat “Insert Operation to perform on each group”
        • “Type”: Sample Standard deviation
        • “On column”: c1
      • param-repeat “Insert Operation to perform on each group”
        • “Type”: Mean
        • “On column”: c2
      • param-repeat “Insert Operation to perform on each group”
        • “Type”: Sample Standard deviation
        • “On column”: c2
      • param-repeat “Insert Operation to perform on each group”
        • “Type”: Mean
        • “On column”: c3
      • param-repeat “Insert Operation to perform on each group”
        • “Type”: Sample Standard deviation
        • “On column”: c3
      • param-repeat “Insert Operation to perform on each group”
        • “Type”: Mean
        • “On column”: c4
      • param-repeat “Insert Operation to perform on each group”
        • “Type”: Sample Standard deviation
        • “On column”: c4
  2. Rename the dataset to iris summary and statistics

    • Click on the galaxy-pencil pencil icon for the dataset to edit its attributes
    • In the central panel, change the Name field
    • Click the Save button
  3. View galaxy-eye the generated file
Question
  1. Can we differentiate the different Iris flower species?
  1. From the results, we can see that the average Iris setosa petal length is lower than 1.5 with a relatively small standard deviation (<0.2). The same can be observed for Iris setosa petal widths. These numbers are much smaller (width and length) than Iris versicolor and Iris virginica petals. We can then use these characteristics to differentiate Iris setosa from the two other species (I. versicolor and I. virginica). On the other hand, we cannot easily differentiate Iris Versicolor from Iris Virginica. Further analysis is necessary.

Visualize Iris dataset features with two-dimensional scatterplots

Let’s visualize the Iris dataset to see how the features depend on each other, and check whether we can spot any immediate patterns.

Hands-on: Plot iris feature pairs in two dimensions
  1. Scatterplot w ggplot2 Tool: toolshed.g2.bx.psu.edu/repos/iuc/ggplot2_point/ggplot2_point/3.3.5+galaxy0 with the following parameters:
    • param-file “Input tabular dataset”: iris clean
    • “Column to plot on x-axis”: 1
    • “Column to plot on y-axis”: 2
    • “Plot title”: Sepal length as a function of sepal width
    • “Label for x axis”: Sepal length
    • “Label for y axis”: Sepal width
    • In “Advanced Options”:
      • “Data point options”: User defined point options
        • “relative size of points”: 2.0
      • “Plotting multiple groups”: Plot multiple groups of data on one plot
        • “column differentiating the different groups”: 5
        • “Color schemes to differentiate your groups”: Set 2 - predefined color pallete
    • In “Output Options”:
      • Additional output format: PDF
  2. View galaxy-eye the resulting plot:

    Contents of the `Group` output dataset.

  3. Rename the dataset to iris sepal scatterplot
Question
  1. What does this scatter plot tell us about Iris species?
  2. Make a new scatter plot, this time with Petal length versus Petal width.
  3. Can we differentiate between the three Iris species?

    Instead of clicking on Scatterplot w ggplot2 tool again, it is possible to recall the previous scatterplot parameters by clicking on re-run button and updating the parameters we wish to modify.

    Hands-on: Re-run the tool
    1. Click on the galaxy-refresh icon (Run this job again) for the output dataset of Scatterplot w ggplot2 tool rerun.

    This brings up the tool interface in the central panel with the parameters set to the values used previously to generate this dataset.

  1. We get similar results than with Summary and statistics: Iris setosa can clearly be distinguished from Iris versicolor and Iris virginica. We can also see that sepal width and length are not sufficient features to differentiate Iris versicolor from Iris virginica.
  2. Scatterplot w ggplot2 Tool: toolshed.g2.bx.psu.edu/repos/iuc/ggplot2_point/ggplot2_point/3.3.5+galaxy0 with the following parameters:
    • param-file “Input tabular dataset”: iris clean
    • “Column to plot on x-axis”: 3
    • “Column to plot on y-axis”: 4
    • “Plot title”: Petal length as a function of petal width
    • “Label for x axis”: Petal length
    • “Label for y axis”: Petal width
    • In “Advanced Options”:
      • “Data point options”: User defined point options
        • “relative size of points”: 2.0
      • “Plotting multiple groups”: Plot multiple groups of data on one plot
        • “column differentiating the different groups”: 5
        • “Color schemes to differentiate your groups”: Set 2 - predefined color pallete
  3. Your new output dataset will look something like this:

Contents of the `Group` output dataset.

We can better differentiate between the 3 Iris species but for some samples the petal length versus width is still insufficient to differentiate Iris versicolor from Iris virginica. And as before, Iris setosa can easily be distinguished from the two other species.

Galaxy management

Convert your analysis history into a workflow

When you look carefully at your history, you can see that it contains all the steps of our analysis, from the beginning to the end. By building this history we have actually built a complete record of our analysis with Galaxy preserving all parameter settings applied at every step. But when you receive new data, or a new report is requested, it would be tedious to do each step over again. Wouldn’t it be nice to just convert this history into a workflow that we will be able to execute again and again?

Galaxy makes this very easy with the Extract workflow option. This means any time you want to build a workflow, you can just perform the steps once manually, and then convert it to a workflow, so that next time it will be a lot less work to do the same analysis.

Hands-on: Extract workflow
  1. Clean up your history: remove any failed (red) jobs from your history by clicking on the galaxy-cross button.

    This will make the creation of the workflow easier.

  2. Click on galaxy-gear (History options) at the top of your history panel and select Extract workflow.

    `Extract Workflow` entry in the history options menu.

    The central panel will show the content of the history in reverse order (oldest on top), and you will be able to choose which steps to include in the workflow.

    Selection of steps for `Extract Workflow` from history.

  3. Replace the Workflow name to something more descriptive, for example: Exploring Iris dataset with statistics and scatterplots.

  4. If there are any steps that shouldn’t be included in the workflow, you can uncheck them in the first column of boxes.

  5. Click on the Create Workflow button near the top.

    You will get a message that the workflow was created. But where did it go?

    `Where workflows go` list.

  6. Click on Workflow in the top menu of Galaxy.

    • Here you have a list of all your workflows.
    • Your newly created workflow should be listed at the top:

    `Your workflows` list.

The workflow editor

Comment: Tip: Problems creating your workflow?

If you had problems extracting your workflow in the previous step, we provide a working copy for you here, which you can import to Galaxy and use for the next sections (see below how to import a workflow to Galaxy).

  • Click on Workflow on the top menu bar of Galaxy. You will see a list of all your workflows.
  • Click on the upload icon galaxy-upload at the top-right of the screen
  • Provide your workflow
    • Option 1: Paste the URL of the workflow into the box labelled “Archived Workflow URL”
    • Option 2: Upload the workflow file in the box labelled “Archived Workflow File”
  • Click the Import workflow button

We can examine the workflow in Galaxy’s workflow editor. Here you can view/change the parameter settings of each step, add and remove tools, and connect an output from one tool to the input of another, all in an easy and graphical manner. You can also use this editor to build workflows from scratch.

Hands-on: Editing our workflow
  1. Open the workflow editor
    • Click on the dropdown menu galaxy-dropdown (triangle icon) to the right of your workflow name.
  2. Select Edit to launch the workflow editor.
    • You should see something like this:

    Workflow editor.

    • When you click on a workflow step, you will get a view of all the parameter settings for that tool on the right-hand side of your screen (the Details section)
    • You can also change the parameter settings of your workflow here, and also do more advanced configuration.
  3. Hiding intermediate outputs
    • We can tell Galaxy which outputs of a workflow are important and should be shown in our history when we run it, and which can be hidden.
    • By default, all outputs will be shown
    • Click the checkbox next to the outputs to mark them as important:
      • outfile in Unique tool
      • out_file1 in Group tool step
        • This should be the Group tool where we performed the counting, you can check which one that is by clicking on it and looking at the parameter settings in the Details box on the right.
      • png in both Scatterplot w ggplot2 tool steps
    • Now, when we run the workflow, we will only see these final outputs
      • i.e. the two dataset with species, the dataset with number of samples by species and the two scatterplots.

    When a workflow is executed, the user is usually primarily interested in the final product and not in all intermediate steps.By default all the outputs of a workflow will be shown, but we can explicitly tell Galaxy which outputs to show and which to hide for a given workflow.This behaviour is controlled by the little checkbox in front of every output dataset:

    Asterisk for `out_file1` in the `Select First` tool

  4. Renaming output datasets
    • When we performed the analysis manually, we often renamed output datasets to something more meaningful
    • We can do the same in a workflow (see the tip box below)
    • Let’s rename the outputs we marked as important with the checkbox (and thus do not hide) to more meaningful names:
      • Unique tool, output outfile: rename to categories tool
      • Group tool, output out_file1: rename to samples per category
      • Rename the scatterplot outputs as well, remember to choose a generic name, since we can now also run this on data other than iris plants.
    1. Open the workflow editor
    2. Click on the tool in the workflow to get the details of the tool on the right-hand side of the screen.
    3. Scroll down to the Configure Output section of your desired parameter, and click it to expand it.
      • Under Rename dataset, give it a meaningful name

        Rename output datasets

  5. Save your workflow (important!) by clicking on the galaxy-save icon at the top right of the screen.

  6. Return to the analysis view by clicking on the Home icon galaxy-home (or Analyze Data on older Galaxy versions) at the top menu bar.
Comment

We could validate our newly built workflow by running it on the same input datasets that we used at the start of this tutorial, in order to make sure we do obtain the same results.

Run workflow on different data

Now that we have built our workflow, let’s use it on some different data. For example, let us explore the diamonds R dataset with it.

Hands-on: Create a new history and upload a new data
  1. Create a new history and give it a name.

    Click the new-history icon at the top of the history panel.

    If the new-history is missing:

    1. Click on the galaxy-gear icon (History options) on the top of the history panel
    2. Select the option Create New from the menu
  2. Import Tool: upload1 the file diamonds.csv from Zenodo or from the data library (ask your instructor)

    https://zenodo.org/record/3540705/files/diamonds.csv
    
    • Copy the link location
    • Open the Galaxy Upload Manager (galaxy-upload on the top-right of the tool panel)

    • Select Paste/Fetch Data
    • Paste the link into the text field

    • Press Start

    • Close the window

    As an alternative to uploading the data from a URL or your computer, the files may also have been made available from a shared data library:

    • Go into Shared data (top panel) then Data libraries
    • Navigate to the correct folder as indicated by your instructor
    • Select the desired files
    • Click on the To History button near the top and select as Datasets from the dropdown menu
    • In the pop-up window, select the history you want to import the files to (or create a new one)
    • Click on Import
  3. Rename galaxy-pencil the dataset to diamonds

    • Click on the galaxy-pencil pencil icon for the dataset to edit its attributes
    • In the central panel, change the Name field
    • Click the Save button
  4. Add a propagating tag galaxy-tags (e.g. #diamonds)

    • Click on the dataset
    • Click on galaxy-tags Edit dataset tags
    • Add a tag starting with #

      Tags starting with # will be automatically propagated to the outputs of tools using this dataset.

    • Check that the tag is appearing below the dataset name

The diamonds dataset comes from the well-known ggplot2 package developed by Hadley Wickham and was initially collected from the Diamond Search Engine in 2008. The original dataset consists of 53940 specimen of diamonds, for which it lists the prices and various properties. For this training, we have created a simpler dataset from the original, in which only the five columns relating to the price and the so-called 4 Cs (carat, cut, color and clarity) of diamond characteristics have been retained.

Comment: The 4 Cs of diamond grading

According to the GIA’s (Gemological Institute of America) diamond grading system

  • Carat refers to the weight of the diamond when measured on a scale
  • Cut refers to the quality of the cut and can take the grades Fair, Good, Very Good, Premium and Ideal
  • Color describes the overall tint, or lack thereof, of the diamond from colorless/white to yellow and is given on a letter scale ranging from D to Z (D being the best, known as colorless).
  • Clarity describes the amount and location of naturally occuring “inclusions” found in nearly all diamonds on a scale of eleven grades ranging from Flawless (the ideal situation) to I3 (Included level 3, the worst quality).

As a further simplification, our training dataset has the qualities in the color and clarity columns re-encoded as integer values (1-23 for color qualities D-Z, and 1-11 for the clarity levels from Flawless to I3). With this adjustment, we can reuse our workflow on the data, and analyze and visualize it following the same steps as we took for the Iris dataset.

Hands-on: Run workflow

To analyze the diamonds price/4 Cs dataset by reusing our workflow:

  1. Open the workflow menu (top menu bar).
    • Find the workflow you made in the previous section,
    • Select the option Run.
    • The central panel will change to allow you to configure and launch the workflow.
  2. Select the diamonds dataset as the input dataset.

  3. Customize the first scatter plot:

    This step is preconfigured to plot column 1 along the x and column 2 along the y axis, while grouping by column 5. This is fine and will result in price getting plotted against carat with grouping by cut, but you would want to adjust the plot title and axis labels accordingly:

    • Change “Plot title” to Diamond price as a function of carat with cut as a factor
    • Change “Label for x axis” to Weight of the diamond (carat)
    • Change “Label for y axis” to Price (US dollars)

    Customize scatter plot.

  4. Customize the second scatter plot.

    This one is preconfigured to plot column 3 along the x and column 4 along the y axis, which, for our new data, would plot color as a function of clarity. However, we would rather want to stick to plotting price against weight in carat as in the first plot, but group by clarity instead of by cut this time, so:

    • Change “Column to plot on x-axis” to 1
    • Change “Column to plot on y-axis” to 2
    • Change “Plot title” to Diamond price as a function of carat with clarity as a factor
    • Change “Label for x axis” to Weight of the diamond (carat)
    • Change “Label for y axis” to Price (US dollars)
    • And finally in “Advanced Options” change “column differentiating the different groups” to 4 (clarity).
  5. Click Run workflow.

  6. Once the workflow has started, you will initially be able to see all its steps, but the unimportant intermediates will disappear after they complete successfully:

    Question
    1. How many cut category are there in the Diamond dataset ?
    2. How many samples are there in each cut category ?
    3. What do you notice about the relationship between price and carat ?
    4. Based on the plot showing Price vs. Carat with Clarity as a factor, do you think clarity accounts for some of the variance in price? Why ?
    1. There are 5 different cut categories:
      • Fair
      • Good
      • Ideal
      • Premium
      • Very Good
    2. We have the following number of samples in each cut category:

      1 2
      Fair 1610
      Good 4906
      Ideal 21551
      Premium 13791
      Very Good 12082
    3. Using any of the scatter plots we made, we can see an obvious positive (non-linear) relationship between both variables: as carat size increases, price also increases. There is also very clear discrete values that carat size takes on, which are those vertical strips on the graph.

      Price by Carat and Clarity.

    4. Holding carat weight constant, we see on the scatter plot shown above that diamonds with lower clarity are almost always cheaper than diamonds with better clarity: diamonds that are “Internally Flawless” are the most expensive whereas “I1” are the least expensive clarity types. So clarity explains a lot of the variance found in price!

Share your work

One of the most important features of Galaxy comes at the end of an analysis. When you have published striking findings, it is important that other researchers are able to reproduce your in-silico experiment. Galaxy enables users to easily share their workflows and histories with others.

To share a history, click on the galaxy-gear icon in the history panel and select Share or Publish. On this page you can do 3 things:

  1. Make History Accessible via Link. This generates a link that you can give out to others. Anybody with this link will be able to view your history.
  2. Make History Accessible and Publish. This will not only create a link, but will also publish your history. This means your history will be listed under Shared Data → Histories in the top menu.
  3. Share with a user. This will share the history only with specific users on the Galaxy instance.
Comment: Permissions

Different servers have different default permission settings. Some servers create all of your datasets completely private to you, while others make them accessible if you know the secret ID.

Be sure to select Also make all objects within the History accessible whenever you make a history accessible via link, otherwise whomever you send your link to might not be able to see your history.

Hands-on: Share history
  1. Share your history with your neighbour.
  2. Find the history shared by your neighbour. Histories shared with specific users can be accessed by those users under their top masthead “User” menu under Histories shared with me.

Conclusion

trophy Well done! You have just performed your first analysis in Galaxy. Additionally you can share your results and methods with others.

Key points
  • Galaxy provides an easy-to-use graphical user interface for often complex command-line tools

  • Galaxy keeps a full record of your analysis in a history

  • Workflows enable you to repeat your analysis on different data

  • Galaxy can connect to external sources for data import and visualization purposes

  • Galaxy provides ways to share your results and methods with others

Frequently Asked Questions

Have questions about this tutorial? Check out the tutorial FAQ page or the FAQ page for the Introduction to Galaxy Analyses topic to see if your question is listed there. If not, please ask your question on the GTN Gitter Channel or the Galaxy Help Forum

References

  1. Fisher, R. A., 1936 The use of multiple measurements in taxonomic problems. Annals of Eugenics 7: 179–188. 10.1111/j.1469-1809.1936.tb02137.x https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1469-1809.1936.tb02137.x

Feedback

Did you use this material as an instructor? Feel free to give us feedback on how it went.
Did you use this material as a learner or student? Click the form below to leave feedback.

Click here to load Google feedback frame

Citing this Tutorial

  1. Anne Fouilloux, Nadia Goué, Christopher Barnett, Michele Maroni, Olha Nahorna, Dave Clements, Saskia Hiltemann, 2022 Galaxy 101 for everyone (Galaxy Training Materials). https://training.galaxyproject.org/training-material/topics/introduction/tutorials/galaxy-intro-101-everyone/tutorial.html Online; accessed TODAY
  2. Batut et al., 2018 Community-Driven Data Analysis Training for Biology Cell Systems 10.1016/j.cels.2018.05.012


@misc{introduction-galaxy-intro-101-everyone,
author = "Anne Fouilloux and Nadia Goué and Christopher Barnett and Michele Maroni and Olha Nahorna and Dave Clements and Saskia Hiltemann",
title = "Galaxy 101 for everyone (Galaxy Training Materials)",
year = "2022",
month = "10",
day = "18"
url = "\url{https://training.galaxyproject.org/training-material/topics/introduction/tutorials/galaxy-intro-101-everyone/tutorial.html}",
note = "[Online; accessed TODAY]"
}
@article{Batut_2018,
    doi = {10.1016/j.cels.2018.05.012},
    url = {https://doi.org/10.1016%2Fj.cels.2018.05.012},
    year = 2018,
    month = {jun},
    publisher = {Elsevier {BV}},
    volume = {6},
    number = {6},
    pages = {752--758.e1},
    author = {B{\'{e}}r{\'{e}}nice Batut and Saskia Hiltemann and Andrea Bagnacani and Dannon Baker and Vivek Bhardwaj and Clemens Blank and Anthony Bretaudeau and Loraine Brillet-Gu{\'{e}}guen and Martin {\v{C}}ech and John Chilton and Dave Clements and Olivia Doppelt-Azeroual and Anika Erxleben and Mallory Ann Freeberg and Simon Gladman and Youri Hoogstrate and Hans-Rudolf Hotz and Torsten Houwaart and Pratik Jagtap and Delphine Larivi{\`{e}}re and Gildas Le Corguill{\'{e}} and Thomas Manke and Fabien Mareuil and Fidel Ram{\'{\i}}rez and Devon Ryan and Florian Christoph Sigloch and Nicola Soranzo and Joachim Wolff and Pavankumar Videm and Markus Wolfien and Aisanjiang Wubuli and Dilmurat Yusuf and James Taylor and Rolf Backofen and Anton Nekrutenko and Björn Grüning},
    title = {Community-Driven Data Analysis Training for Biology},
    journal = {Cell Systems}
}
                   

Congratulations on successfully completing this tutorial!