How to create a basic GUI for rna seq pipeline

Question

[pubsurfted]

Hello everyone.

I'm a bio core student who needs guidance regarding which tools to use and where.

I have a project in which I want to build a website with basic GUI that will give the user options for each rnaseq pipeline step, i.e. for quality control trimming the user can select the trimmer of their choice (trimmomatic, trimgalore, cutadapt) from a dropdown menu.

So my question is how do I go about producing this? Should I write scripts in nextflow as backend and use react as frontend?

Please excuse if this is a really dumb question. I'd appreciate your kindness.

[Pierre Lindenbaum]

have a look at zenity, or better, tell your user how to use a CLI.

[Joe]

Is there anything you need this site to do that something like Galaxy doesn't already do?

[young_bioinformatician]

I would firstly write a pipeline script using shell script, CWL, Nextflow, or any language., and then create a PHP server and use it to make GUI. I have done it before for Trinity tool in my undergraduate years, but of course It was very basic.

In the end, It will be a web-based GUI tool.

Answer 1

[Marcel Ribeiro-Dantas]

You're probably looking for something like nf-core launch. Check here.

It allows you to launch an nf-core [compatible] pipeline from your web browser, either starting from the command line or not. You get a form to fill the arguments/options of your pipeline and so on. If you want something even better than that, you should try Nextflow Tower.

[sbt_gvs]

I have a Pipeline in WDL , that runs some bash and R processing scripts on remote supercomputers. Is there any UI available for WDL pipeline? I began building one using Rshiny, but seems more complicated than perceived. What is the best approach you would suggest ? Should I change to Nextflow and use the above recommended or is there any UI for snakemake that would enable non-programmers to run pipeline easily ?

[Marcel Ribeiro-Dantas]

I'm sorry, I don't know much about WDL :\ I think switching to Nextflow and using nf-core launch or Nextflow Tower is the better option here, but I'm biased as I work for Seqera Labs and with Nextflow.

Answer 2

[Matthias Zepper]

Please do not reinvent the wheel - for your own sanity, because if you want to write that from the bottom up, you need a team of software engineers and at least (!) a year of time*.

So rather check what is already out there in the Bioinfo world: Galaxy was already mentioned, also Terra/AnVIL might be mature by now. When it comes to the workflow languages that are commonly used in bioinformatics (Nextflow, Snakemake, CWL/WDL) most other UIs are commercial offerings: Tower for Nextflow and the SevenBridges platform is CWL-based. I currently don't recall what DNAnexus, Paradigm4 and Lifebit run on, but since all of those are commercial, they are probably irrelevant for you anyway.

An alternative approach would be to port an existing pipeline to another workflow DSL, which ships with an open-source UI. I am particularly thinking of Airflow and Dagster/Dagit, but there are literally hundreds of workflow executors and pipeline DSLs out there, so there might be better ones around by now. Yet, since you would need to write each module and take care of deploying the tool's container somewhere, this would still be a massive amount of work. So if this is a classroom assignment, I would ask for a more appropriate task.

*(I speak from some experience here, since I unsuccessfully tried to found a start-up for launching bioinfo pipelines in the cloud with a nice UI in the browser between 2019-2021. The sheer amount of complexity in such a system is staggering when it comes to the nitty-gritty details.)

[Marcel Ribeiro-Dantas]

In general, I agree with you. Just wanted to mention that Tower Cloud is free for most use cases and nf-core tools (launch) is free, open source, and does what he seems to be looking for.

[pubsurfted]

Thank you for taking the time to give such a thorough reply. I am obviously not at the stage to reinvent the wheel and I am not looking to reinvent it. My goal was to learn about the processes and I learn better when I replicate things. I should've mentioned that in my post.