iiinova — where your digital self keeps evolving

Paul Harrison

@paulfharrison.bsky.social

459 followers 128 following 101 posts

Bioinformatician at Monash University, Melbourne, Australia. I also use mastodon: @pfh@mastondon.online mastodon.online/@pfh My homepage is: logarithmic.net/pfh On Twitter I was: @paulfharrison

Posts Replies Media

Paul Harrison @paulfharrison.bsky.social · 27/07/2025

uvx demakein I've finally updated my wind instrument design program to Python 3. It only took me 10 years to get around to. I was pleased to find there is now a fairly solid python library for 3D boolean operations (manifold3d). github.com/pfh/demakein

A picture of a 3D printed whistle in front of a 3D printer.

230

Paul Harrison @paulfharrison.bsky.social · 18/04/2025

Some conventional flow matching:

A computer generated image that looks like daubs of colored oil-paint.

020

Paul Harrison @paulfharrison.bsky.social · 09/03/2025

Happy little accidents...

140

Paul Harrison @paulfharrison.bsky.social · 08/03/2025

I'm really liking this course on generative diffusion models. They seem to have boiled many years of confusing development of ideas down to a simple approach. diffusion.csail.mit.edu

Generation of samples can be adjusted to output samples that are very clearly distinguishable as members of the desired class.

050

Paul Harrison @paulfharrison.bsky.social · 23/02/2025

Second, sampling from the distribution with a Langevin Dynamics simulation. The algorithm is almost identical to gradient descent with momentum, but we add just the right amount of noise to the momentum at each step.

110

Paul Harrison @paulfharrison.bsky.social · 23/02/2025

Comparison of optimization and sampling from a distribution defined by an energy function. I use a continuous version of the Ising model spin lattice energy. First, optimization from a random initial state using gradient descent with momentum, using the SGD optimizer in PyTorch.

120

Paul Harrison @paulfharrison.bsky.social · 05/12/2024

Here's a new version of the plot. There is one point per gene! The y axis shows the estimated log fold change, and the color tells about the confidence bound. I lose a little resolution by using color, but hopefully gain understandability. I am hoping it is less confusing and more conventional.

110

Paul Harrison @paulfharrison.bsky.social · 05/12/2024

Here's the plot. It's looking at differential gene expression. There are two type of points. They gray dots show estimated log fold change on the y-axis. The colored points show a confidence bound on the log fold change on the y-axis. A significant gene is represented by two different points!

200

Paul Harrison @paulfharrison.bsky.social · 27/11/2024

This was one of the important ideas:

010

Paul Harrison @paulfharrison.bsky.social · 03/10/2024

This week the Monash Genomics and Bioinformatics Platform did a bulk RNA-Seq workshop, covering end-to-end from experimental design, through library preparations, running an analysis pipeline, and digging into differential expression.

100

Paul Harrison @paulfharrison.bsky.social · 29/12/2023

I made a short video of the strange things UMAP and t-SNE can do to your data. The algorithms are shown mostly working as intended, yet with some surprising consequences. #UMAP #tSNE #scRNAseq #wtf www.youtube.com/watch?v=gwqU...

030