Lightnews — Scholar-powered news

Ming Tommy Tang @tommytang.bsky.social · 18h

The genomic landscape of relapsed infant and childhood KMT2A-rearranged acute leukemia www.nature.com/articles/s4...

Ming Tommy Tang @tommytang.bsky.social · 19h

I hope you've found this post helpful.

Follow me for more.

Subscribe to my FREE newsletter chatomics to learn bioinformatics divingintogeneticsandgenomics.ck.page/profile

Hi! I'm Tommy Tang

I am a bioinformatician/computational biologist with six years of wet lab experience and over 12 years of computation experience. I will help you to learn computational skills to tame astronomical data and derive insights. Check out the resources I offer below and sign up for my newsletter!

divingintogeneticsandgenomics.ck.page

Ming Tommy Tang @tommytang.bsky.social · 19h

14/14
Bioinformatics isn’t just code.
It’s systems engineering in disguise.
Every error conquered = deeper understanding.
You’ve got this. 💪

1

Ming Tommy Tang @tommytang.bsky.social · 19h

13/14
Key Takeaways:
Red text = clues, not failure

90% of errors are missing system libs

Isolate issues with clean environments

Document every fix—you’ll forget

1 1

Ming Tommy Tang @tommytang.bsky.social · 19h

12/14
Windows users:
Install Rtools for C/C++ dependencies:
cran.r-project.org/bin/windows...

Restart RStudio after installing!

1

Ming Tommy Tang @tommytang.bsky.social · 19h

11/14
Permission denied?
Never use sudo R. Instead:

install.packages("pkg", lib="~/R/libs") # User directory

Add export R_LIBS_USER=~/R/libs to .bashrc.

1

Ming Tommy Tang @tommytang.bsky.social · 19h

10/14
Nuclear option: Start fresh.

R --vanilla # No startup configs
install.packages("problematic_pkg")

Or use renv/conda for isolated environments.

1

Ming Tommy Tang @tommytang.bsky.social · 19h

9/14
ChatGPT Prompt Template:
"I’m on macOS Ventura, R 4.3.0. Trying to install DESeq2 with BiocManager::install(). Got this error: [paste]. Whats wrong?
Context matters.

1

Ming Tommy Tang @tommytang.bsky.social · 19h

8/14
Still stuck?
Google the exact error in quotes:
"error: X11 headers not found" site:stackoverflow.com (you search with that site! cool google trick).
Add R or Bioconductor to narrow results.

Newest Questions

Stack Overflow | The World’s Largest Online Community for Developers

stackoverflow.com

1 1

Ming Tommy Tang @tommytang.bsky.social · 19h

7/14
Bioconductor packages: Always use:
BiocManager::install("limma") # Not install.packages()!

Bioconductor has its own dependency tree.

1 1

Ming Tommy Tang @tommytang.bsky.social · 19h

6/14
Installing Seurat? Common error:
libgfortran.so.3: cannot open...
Linux:
sudo apt install libgfortran3

macOS:
brew install gcc # Installs Fortran libs

1 1

Ming Tommy Tang @tommytang.bsky.social · 19h

5/14
Linux example:
Error: X11 headers not found
Fix:
sudo apt-get install libx11-dev # Debian/Ubuntu

if on macOS
Install XQuartz (X11 server) (This has happened several times to me!)

1

Ming Tommy Tang @tommytang.bsky.social · 19h

4/14
Don’t scroll past the error.
Read line by line.
Look for:
cannot find -lxml2 → Install libxml2-dev

zlib.h not found → Install zlib1g-dev (Linux) / zlib (macOS)

1 1

Ming Tommy Tang @tommytang.bsky.social · 19h

3/14
Most install errors boil down to:
✅ R-level issues: Missing dependencies (e.g., limma)
✅ System-level issues: Missing libraries (e.g., zlib)
✅ Permissions: Can’t write to /usr/lib/

1 1

Ming Tommy Tang @tommytang.bsky.social · 19h

2/14
You run:
install.packages("BiocManager") # For Bioconductor packages

Boom. Red wall. Panic sets in.
But breathe. The clues are there—if you know how to read them.

1 1

Ming Tommy Tang @tommytang.bsky.social · 19h

1/14
Installation fail. Red text. Cryptic errors.
Every bioinformatician has been there—even experts.
It feels personal. It’s not.

1

Ming Tommy Tang @tommytang.bsky.social · 19h

You just want to install an R package.
But the screen goes red.
Your stomach drops.
Even pros dread this moment. Here's how to survive it: 🧵

1 1 3

Ming Tommy Tang @tommytang.bsky.social · 19h

I hope you've found this post helpful.

Follow me for more.

Subscribe to my FREE newsletter chatomics to learn bioinformatics divingintogeneticsandgenomics.ck.page/profile

Hi! I'm Tommy Tang

I am a bioinformatician/computational biologist with six years of wet lab experience and over 12 years of computation experience. I will help you to learn computational skills to tame astronomical data and derive insights. Check out the resources I offer below and sign up for my newsletter!

divingintogeneticsandgenomics.ck.page

Ming Tommy Tang @tommytang.bsky.social · 19h

12/
Bioinformatics isn’t about blindly analyzing big matrices.
It’s about asking:
“What’s the signal?
What’s the noise?
And what can I ignore to see the truth?”
That’s the real art of dimension reduction.

1

Ming Tommy Tang @tommytang.bsky.social · 19h

11/
Takeaways:
High dimensions can lie

Use PCA/UMAP wisely

Interpret results with biology in mind

Don’t trust t-SNE for distances

Feature selection is your first defense

1

Ming Tommy Tang @tommytang.bsky.social · 19h

10/
Bottom line:
Dimensionality is a curse only if you treat all data as equal.
The solution is knowing when to cut, compress, or interpret.

1

Ming Tommy Tang @tommytang.bsky.social · 19h

9/
Key challenges in scRNA-seq:
Sparsity

Batch effects

Scalability
You need tools that know how to swim in noisy waters—Seurat, Scanpy, scVI.

1

Ming Tommy Tang @tommytang.bsky.social · 19h

8/
scRNA-seq dimension reduction steps:
Pick highly variable genes

Use PCA to drop noise

UMAP to visualize clusters

Harmony/scVI to align across batches. (Harmony corrects the PC coordinates)

Autoencoders to denoise

1

Ming Tommy Tang @tommytang.bsky.social · 19h

7/
This works in single-cell RNA-seq too—except now you’ve got 20,000 genes AND 50,000 cells.
Welcome to p ≈ n.
But there’s a twist: the matrix is full of zeroes. That’s dropout.

1

Ming Tommy Tang @tommytang.bsky.social · 19h

6/
What does PCA really do?
Take 20,000 genes and build 10 orthogonal “super-genes” (PCs).
They explain variance, not necessarily biology.
Still, it helps you see patterns you’d never spot in raw data.

1