Frederik Ziebell
@fredonion.bsky.social
Mathematician turned systems Biology; PhD @ DKFZ, PostDoc @ EMBL
When communicating to others that clustering is just exploratory, I highlight that points at the borders of a cluster are often closer to points of another cluster than to their own centroid, so it isn't perfect. Frank Harrell has a good blog post: www.fharrell.com/post/cluster/
The Burden of Demonstrating Statistical Validity of Clusters – Statistical Thinking
Patient clustering, often described as the finding of new phenotypes, is being used with increasing frequency in the medical literature. Most of the applications of clustering of observations are not ...
www.fharrell.com
October 17, 2025 at 5:33 AM
When communicating to others that clustering is just exploratory, I highlight that points at the borders of a cluster are often closer to points of another cluster than to their own centroid, so it isn't perfect. Frank Harrell has a good blog post: www.fharrell.com/post/cluster/
Even better would be to have Quarto notebooks with inline chunk output to have truly interactive multi-language analysis, and I know there are others desparately waiting for this feature: github.com/posit-dev/po...
Add notebook support for Quarto and RMarkdown with Chunk Output inline · Issue #5640 · posit-dev/positron
This is a plea to bring back the amazing notebook support from RStudio. I decided to create an issue despite the advice given here because I only found that discussion thread after I did several ho...
github.com
August 14, 2025 at 8:41 AM
Even better would be to have Quarto notebooks with inline chunk output to have truly interactive multi-language analysis, and I know there are others desparately waiting for this feature: github.com/posit-dev/po...
But not every matrix is a square and for data matrices, rows and columns have different meanings (observations vs. variables).
July 1, 2025 at 8:11 PM
But not every matrix is a square and for data matrices, rows and columns have different meanings (observations vs. variables).
When I skimmed over your paper, I instinctively looked for the figure comparing your package to established ones to decide if I want to try it out, and couldn't find it.
February 6, 2025 at 7:49 PM
When I skimmed over your paper, I instinctively looked for the figure comparing your package to established ones to decide if I want to try it out, and couldn't find it.
What I meant was, that the tables are, in my view, not as quick to comprehend as a simple plot. The aim of a method paper should not be to present information in the most space efficient manner, but to advertise so that others want to use it.
February 6, 2025 at 7:49 PM
What I meant was, that the tables are, in my view, not as quick to comprehend as a simple plot. The aim of a method paper should not be to present information in the most space efficient manner, but to advertise so that others want to use it.
Could you maybe create a few plots for the benchmark section? These bench::mark output tables are difficult to parse. It would be better to visually show how much faster collapse is than competitors.
February 6, 2025 at 2:38 PM
Could you maybe create a few plots for the benchmark section? These bench::mark output tables are difficult to parse. It would be better to visually show how much faster collapse is than competitors.
I can't edit my post, the tool is called Termux: conr.ca/post/install...
January 19, 2024 at 2:49 PM
I can't edit my post, the tool is called Termux: conr.ca/post/install...
You can already do that with tmux. My phone even runs DESeq2. 13.1sec for the airway data vs. 12.7sec on my MacBook pro (single threaded).
January 19, 2024 at 9:38 AM
You can already do that with tmux. My phone even runs DESeq2. 13.1sec for the airway data vs. 12.7sec on my MacBook pro (single threaded).