A simulated annealing algorithm for randomizing weighted networks” | nature.com/articles/s43... How do we randomize brain connectomes (and other networks) with biological edge weights? @richardfbetzel.bsky.social @natcomputsci.bsky.social ⤵️ - ThreadSky

About ThreadSky

misicbata.bsky.social • 73 days ago

A simulated annealing algorithm for randomizing weighted networks” | https://nature.com/articles/s43588-024-00735-z

How do we randomize brain connectomes (and other networks) with biological edge weights? @richardfbetzel.bsky.social
@natcomputsci.bsky.social ⤵️

Comments

misicbata.bsky.social•73 days ago

1️⃣ Statistical inference in connectomics works by comparing networks with randomized "null" models.

But current randomization algorithms were developed for binary networks without edge weights.

Modern imaging tech generates brain networks with meaningful biological weights.

misicbata.bsky.social•73 days ago

2️⃣ Here we introduce and benchmark an algorithm that randomizes weighted networks while selectively preserving size, density, and weighted degree (“strength”) sequence.

Our implementation is based on simulated annealing, a simple and widely-used metaheuristic.

misicbata.bsky.social•73 days ago

3️⃣ The simulated annealing algorithm outperforms existing methods across multiple scenarios.

misicbata.bsky.social•73 days ago

4️⃣ Using the more accurate simulated annealing algorithm leads to fundamentally different inference about established findings, such as the rich-club phenomenon.

misicbata.bsky.social•73 days ago

5️⃣ The algorithm can be readily extended to accommodate different types of networks, including directed networks from tract tracing and genetic labeling.

misicbata.bsky.social•73 days ago

6️⃣ The algorithm can be extended to signed networks, such as functional connectivity (fMRI), metabolic connectivity (PET), ephys connectivity (MEG) or correlated gene expression.

Posting Rules

Be respectful to others
No spam or self-promotion
Stay on topic
Follow Bluesky's terms of service

Comments

Posting Rules

Reply