DIY with own (or other) data
1 Introduction
During this session, you will have the opportunity to work by yourself with the tools that we have introduced during the workshop.
The instructors will be available to help you with any questions you may have.
You can use your own data if you’d like. Let us know if you need any help with getting started, such as uploading your data.
You can also continue working on the example data. We have a bit more info on that below.
We also skipped over a number of exercises in the previous sessions. Feel free to go back and try those out, if you want to.
Remember that you can also ask AI —including the Gemini options Jonathan showed— to help brainstorm what to do! Use this as an opportunity to explore the possibilities of AI in your own research context.
Whichever data set you use, perhaps a nice end goal is to create one or more plots, i.e. visualizations of your results. Even if you, say, only align reads to a reference genome, you can still make a plot of the alignment success statistics.
Don’t go overboard with very big asks, especially ones far removed from research data analysis context that is the focus of the this workshop. For example, we don’t really want to you to build a massive web app for personal use, here. If you want to try something along those lines, please use your own API key.
If you’re in doubt, please ask the instructors for guidance.
2 Working with the example dataset
Below are some suggestions if you’d like to continue working on the Iqbal et al. (2025) dataset.
A core first goal of almost any RNA-Seq analysis is to produce a “count table”, which has counts for each gene (a proxy for gene expression levels) for each sample. You can try to produce such a count table. If you’re unsure about the steps needed to do this, you can start by asking AI! (Your
CLAUDE.mdmay already have some instructions for this.)This can be done step-by-step with separate shell scripts, or for more advanced (or ambitious!) users, you can try to write a simple but reproducible Nextflow pipeline.
If you’re going to submit a bunch of Slurm jobs, including with
forloops to loop over files or samples, you’ll also want a separate file (shell script, or optionally Markdown file with code blocks) to store that code. As with everything else, you can ask AI to help you write that code and design the way to organize scripts.If you would prefer to start with more “downstream” analyses on the count table, such as making a PCA plot of expression profiles, doing differential expression analysis, making heatmaps, and so on.
There is a pre-made count table at
/fs/scratch/PAS3454/share/iqbal_results/salmon.merged.gene_counts_length_scaled.tsvNoteNot sure what to do, exactly?One option is to give Claude some context and ask it what to do!
You can also take a look at this tutorial page which goes over the typical steps of an RNA-Seq analysis: RNA-Seq analysis tutorial. This is a different dataset, so don’t try to follow the exact code — use Claude to adapt it, or start from scratch.
NoteWorking with R? Consider setting up and using PositronSee our Positron setup page!
Whole-genome resequencing data
If you’d like to work with the whole-genome resequencing data from this paper instead of the RNA-Seq data, this is available in /fs/scratch/PAS3454/data/wgs_iqbal.