5 Must Dos For Efficient Bioinformatics
Aug 3, 2017
5 must-dos for efficient bioinformatics
My colleague Kamil was joking about how we need to make bioinformatics sexier and more click-bait-y with those ridiculous buzz-feed-esque headlines like ‘N ways to X your Y’ and ‘M best Ws that will K your J’. So here are my 5 must-dos for efficient bioinformatics that I’ve tried to get all my students to adopt.
-
Get a text editor that can send commands to the terminal
The biggest efficiency killer I’ve noticed from observing my students is how they write code in a text editor and then copy and paste that code to execute it in an interactive session. Then they’ll have a bug in the code, fix it in their interactive session, but then have to remember to copy it back to their text editor. What? Time is wasted in just highlighting code with the mouse, copying and pasting…not to mention the flurry of missed tabs or semicolons from copy-and-paste errors. Using an IDE like RStudio for R and Spyder for Python is a good way to mitigate these issues. My preference is to use Emacs with ESS. Whatever your preference, just find a way to write code in an editor and then execute that code directly with a keyboard shortcut instead of copying and pasting. -
Learn keyboard shortcuts
Speaking of keyboard shortcuts, this is necessary when using text editors like Emacs and Vim but can also be extended to your browser with plug-ins like Vimium. When you work on the computer for as many hours as I do, you’ll likely start developing muscle fatigue from large movements such as from the keyboard to the mouse and vice versa. So I try to use the mouse as little as possible by taking advantage of keyboard shortcuts to do everything from clicking on links or executing large chunks of code. The learning curve can be a bit steep and, depending on your level of discipline, it’s very tempting to just use the mouse. So I recommend my students remove their mouse for a week to force themselves to learn keyboard shortcuts (easier said than done but the recommendation is made). -
Use a window manager
There are many options for window managers but I prefer Screen or Tmux. By using a window manager, you can multiplex your window and see how much compute, memory, etc you’re using. But, my students will say, you could always just open up 3 separate windows and tile them next to each other… STILL, the advantage of a window manager I see is for when you have a crappy connection. With a window manager, if your connection dies, your session can still be running and you can just reattach to it when you get your connection back. Plus, with a window manager like screen, I can log on to the server from home, start a screen session, begin running some stuff, close my computer, bus to work, and then reattach to that session once I’m at work. And by then, whatever I started running from home will be finished! -
Adopt a workflow management system
Again, there are many, many, many options for workflow management systems but I like to use Snakemake. Snakemake workflows are essentially Python scripts extended by declarative code to define rules. For example, I have a ‘rule’ for converting from SRA to fastqs, another rule for aligning using TopHat, another rule for aligning using STAR, another rule for marking duplicates using picard, another rule for quantifying gene expression using featureCounts, another ‘rule’ for using Kallisto, another rule for variant calling using GATK, yada yada yada. Then I can piece together these rules to create a uniform workflow that can be applied to lots of different datasets. -
DOCUMENT DOCUMENT DOCUMENT!
If I read through your code, given enough time, I can probably figure out what it does. But I can understand it much faster, if you just tell me what it does. Likewise, 3 months from now, future-you will be able to understand what past-you did a lot faster if current-you just documented properly. If you’ve written a function, write a line about what it does. What are the parameters? What is the output? I like to document as roxygen comments. This will save you time later when you decide to package everything together and write the package documentation.
There are lots of other best practices I recommend for my students including following certain style guides to keep code legible, version controlling using git or svn, and organizing folders in a particular way…but I will save that list for another post ;)
What are your tricks for improving efficiency in bioinformatics, coding, and beyond?
- Older
- Newer
RECENT POSTS
- The many ways to calculate Moran's I for identifying spatially variable genes in spatial transcriptomics data on 29 August 2024
- Characterizing spatial heterogeneity using spatial bootstrapping with SEraster on 23 July 2024
- I use R to (try to) figure out which hospital I should go to for shoppable medical services by comparing costs through analyzing Hospital Price Transparency data on 22 April 2024
- Cross modality image alignment at single cell resolution with STalign on 11 April 2024
- Spatial Transcriptomics Analysis Of Xenium Lymph Node on 24 March 2024
- Querying Google Scholar with Rvest on 18 March 2024
- Alignment of Xenium and Visium spatial transcriptomics data using STalign on 27 December 2023
- Aligning 10X Visium spatial transcriptomics datasets using STalign with Reticulate in R on 05 November 2023
- Aligning single-cell spatial transcriptomics datasets simulated with non-linear disortions on 20 August 2023
- 10x Visium spatial transcriptomics data analysis with STdeconvolve in R on 29 May 2023