Skip to content

Software Installation (Conda)

Learning Objectives

By the end of this section, you should be able to:

  • Explain why Conda is useful on shared systems
  • Install Conda in your home directory
  • Create, activate, and deactivate project environments
  • Install software into a specific environment and export it for reproducibility

At the end of the previous section, we tried to run the tree command. Did you get the expected output? No? Well, that is because you do not have the tree command installed.

Since this is a shared computational resource and you do not have administrative rights to install new programs, we have to use a different program called conda which is able to install programs in your home directory so you can run them.

1. Why Use Conda?

Conda allows a user on a Unix system to install packages without sudo/admin privileges. Additionally, Conda allows users to create "environments" where a specific suite of programs are installed. A user can have as many environments as they want (i.e., for different projects and analyses types). This has the benefit of "freezing" your program versions so that your analyses are reproducible.

Imagine running an analysis, getting your final results, then 2 years later, your PI wants you to re-run the analysis with new samples. However, the programs you used were updated many times in the past years (bug fixes, new features, etc). Perhaps the output format has changed and now you can't compare the results of the new dataset with the old dataset. Using a Conda environment will allow you to recapitulate the original analysis for the new dataset.

2. Installing Conda

In order to install Conda, we will need to download the program directly to our remote Linux computer, Pronghorn, or on your local workstation.

Visit this website here: https://github.com/conda-forge/miniforge.

If you scroll towards the bottom, you can see a list of Linux, OSX, and Windows installation options. We will use the one labeled "Linux X86_64". You can right-click and copy the link location from your browser, but I also provided the link below: https://github.com/conda-forge/miniforge/releases/latest.

We will use the wget utility, short for web get, to download this file to Pronghorn.

wget https://yourlinkhere

Next, we will run this file with the program bash in order to install Conda.

bash Mambaforge-24.9.2-0-Linux-x86_64.sh

There will be some prompts about the installation which we will proceed through together. In order to scroll through the EULA agreement, use the space bar.

Reloading Configuration Required

Once done, you will need to reload the .bashrc configuration to get your new conda installation working. You can either restart your terminal or reload your environment. Please do this now after installing conda by typing exit.

Reloading the Environment

Instead of logging completely out and back in by typing exit, you can tell the terminal to refresh its settings immediately. Use the source command on your bash profile:

source ~/.bashrc

After reloading, test to see if Conda is working by using the conda command. If you do not receive a "command not found" error, you successfully installed Conda.

Verify Conda Installation

conda --version
conda info

If these commands run, your installation is working.

3. Creating and Activating Environments

Use named environments to isolate software for each project.

conda create -n unix_workshop python=3.11
conda activate unix_workshop
conda env list

When you are done working:

conda deactivate

4. Installing a Program with Conda

Now that we have conda installed, let's install the tree command. This will install tree into your currently loaded environment, which the default is "base". 

conda install -c conda-forge tree htop

Understanding the Command:

  • conda install: The base command to download and install a new tool.
  • -c conda-forge: Tells Conda to look in the conda-forge "channel" (repository) for the software.
  • tree htop: The specific packages we want to install.

5. Exporting and Recreating Environments

Capture exact versions to make your work reproducible.

conda activate unix_workshop
conda env export > unix_workshop_env.yml

To recreate the same environment on another system:

conda env create -f unix_workshop_env.yml

6. Monitoring Running Jobs

The top command prints a self-updating table of running processes and system stats. Use q to exit top, z to toggle better color contrast, M to sort by memory use, P to sort by processor use, and c to toggle display of the full commands. Hit 1 to toggle display of all processors, and hit u followed by typing in a username in order to only show processes (jobs) owned by that user.

Since we just installed it, let's also compare the new htop command to the system default top command.

HPC Job Schedulers

While htop is an excellent tool for monitoring interactive sessions, keep in mind that you generally won't run resource-heavy bioinformatics tools directly on the login node. For heavy computational lifting, you will eventually package these commands into scripts and submit them to a cluster using job schedulers like SLURM or PBS.

Troubleshooting: Conda Not Found

If conda is still not available after install:

  • Run source ~/.bashrc again.
  • Open a new terminal session and re-test with conda --version.
  • Confirm installation files exist in your home directory.