The guide I wish I had 12 years ago: a step-by-step guide to replicate a genomics paper figure

Unlock your bioinformatics potential! 🌟 Discover why mastering both Python and R can elevate your data visualization and analysis skills. Dive into the world of code! 💻 #Bioinformatics

Join millions who have switched to Grok.

Enjoy this tweet? follow me @tangming2005 and join my newsletter to learn computational biology

High-throughput profiling of chemical-induced gene expression across 93,644 perturbations.

I hope you've found this post helpful. Follow me for more. Subscribe to my FREE newsletter chatomics to learn bioinformatics

It should be a required reading for any scientists. It is probably the simplest thing to do that can have a high ROI.

and I know wet lab scientists are willing to do good work,.they are just not aware of the right way to do it (I was one of them). that's why I want to educate more. DO read this article: Data Organization in spreadsheets.

You can save millions $ for your organization if everyone follows best practices to fill in a spreadsheet. Do simple things to make bioinformatician's life easier. Read the full article here

I hope you've found this post helpful. Follow me for more. Subscribe to my FREE newsletter chatomics to learn bioinformatics

10/. Your turn: what’s the worst metadata bug you’ve faced?. Mine was a cancer subtype column where “Basal” appeared as:. “Basal”, “basal-like”, “Basal_Like”, and “BASAL”. What’s yours?.

9/. Key takeaways:.Metadata is always messier than you think. Standardize early. Inspect everything. Cleaning isn’t busywork—it’s survival.

8/. Every wasted hour I’ve had in bioinformatics traces back to one thing: ignoring data cleaning. Get this step right, and the rest flows.

7/. This is detective work. Not glamorous, but essential. Before PCA or UMAP—real expertise starts with knowing your metadata.

6/. Here’s how I clean them in R:.Use janitor::clean_names() to make columns consistent. Replace text placeholders with true NA using na_if(). Run unique() or table() to spot strange values. janitor::tbly() is your friend too.

5/. If you don’t standardize them, filters fail. Joins drop rows. You’ll think your pipeline is wrong, when in fact your metadata is lying to you.

4/. I’ve seen the same “missing” value represented as:. <NA>. [Not Available]. N/A.They look the same. But to R, they’re not. read more details here.

3/. That was just the start. Metadata in bioinformatics is full of hidden traps. The most dangerous? Missing values.

2/. See the problem?. One has underscores. The other has a dot. This tiny difference is enough to break your code.

1/. I was using TCGAbiolinks to analyze TCGA data. The metadata for transcription subtypes had two column names:. paper_expression_subtype. paper_Expression.Subtype.

Bioinformatics data is messy. Here’s the nightmare that almost broke me 🧵👇

Accurate and scalable multi-disease classification from adaptive immune repertoires.