iiinova — where your digital self keeps evolving

Claudia C. Weber

@cc7740.bsky.social

999 followers 722 following 68 posts

Molecular evolution, genomes (lots of them, preferably weird), computational models. Stockholm, Sweden (she/her) github.com/claudia-c-weber

Posts Replies Media

Claudia C. Weber @cc7740.bsky.social · 27/06/2025

Shiny wasp

1153 0

Claudia C. Weber @cc7740.bsky.social · 18/06/2025

Hymenopteran Hotel

A bee poking its head out of a hole in a log

010 0

Claudia C. Weber @cc7740.bsky.social · 20/04/2025

First cooperative butterfly of the year

Speckled Wood butterfly perching on a twig

0110 0

Claudia C. Weber @cc7740.bsky.social · 04/12/2024

What about two dinos?

A kestrel and a magpie, perched on the stumps of a cut down tree in front of a lake.

151 0

Claudia C. Weber @cc7740.bsky.social · 04/11/2024

The result: One Kudoa genome assembly that is near-chromosomal, and another that is contiguous enough to look at gene order conservation... Data have been submitted to ENA (see github.com/CobiontID/Ku... for more info). If you're wondering what on earth a VAE is: doi.org/10.1093/g3jo...

Chromatin contact map for Kudoa neothunni, showing three inferred chromosomes

110 0

Claudia C. Weber @cc7740.bsky.social · 04/11/2024

To retrieve the rest of the genomes, we went back to the reads. Neither sample contained much parasite sequence (dark colours indicate low coverage). Meta-assembly worked well in one case. In the second case, removing the majority of fish reads based on VAE embeddings made the assembler's job easier

A) VAE embeddings of read tetranucleotide content for the Trachurus trachurus sample, coloured by median 31-mer occurrence across the whole set. Darker colours indicate low k-mer coverage. A cluster of low-coverage myxozoan reads is visible on the right. B) K-mer histograms showing the entire dataset (blue, upper curve), and the selection corresponding to the parasite (green, lower curve)

100 0

Claudia C. Weber @cc7740.bsky.social · 04/11/2024

Identifying the parasite sequences is tricky due to missing reference data. Cross-referencing distinct components of the chromatin interaction network (identified by unsupervised clustering) and ribosomal markers provided an initial set of myxozoan scaffolds.

A network graph showing the chromatin contact network of the Thunnus albacares sample, with labels based on unsupervised clustering of the chromatin interaction network. Yellow points indicate scaffolds containing Kudoa small ribosomal subunit sequences. Fish sequences (blue) are shown on the left. Parasite sequences (green) are shown on the right.

100 0

Claudia C. Weber @cc7740.bsky.social · 09/10/2024

Spider!

030 0

Claudia C. Weber @cc7740.bsky.social · 04/06/2024

Given compositional differences, sequences from different sources form distinct clusters in the VAE’s latent space. For example, the pictured moth (blue) is infected with Wolbachia (red). No labels? Discrepancies in coverage and coding density (second image) often provide clues.

Scatter plot of the first and second latent dimensions of a VAE for reads from the buff tip moth. Reads that map to Wolbachia for a distinct cluster, and are shown in red.

Scatter plot of the first and second latent dimensions of a VAE for reads from the buff tip moth, coloured by estimated coding density (highest density: yellow). The highlighted Wolbachia sequences show the highest density.

100 0

Claudia C. Weber @cc7740.bsky.social · 01/12/2023

Colourful waxcaps

Bright red waxcap fungi framed by green grass

000 0

Claudia C. Weber @cc7740.bsky.social · 03/10/2023

I've had this as a desktop background for eons (Wissahickon Creek in early November)

A photo of ducks swimming on water reflecting autumn foliage

130 0