This week’s recap highlights new methods in genetic epidemiology, mostly centered around genomic data sharing and privacy-preserving methods: a short commentary on genomic data sharing highlighting how new challenges complicate large-scale data sharing practices, a privacy-preserving method for QTL mapping, privacy-preserving methods for federated biobank-scale GWAS analysis, a Nextflow pipeline for polygenic score QC and construction, and new

I realize that the titular statement is open to misinterpretation so let me head that off at the pass: I’m not saying this prescriptively, like you should learn anatomy to become a better person (you should learn anatomy because it’s accessible and it rules), or that knowing anatomy makes people better. I’m also not saying this distributively, like anatomists are better people than non-anatomists.

How QwQ-32B Is Redefining Efficiency and Reasoning in Open-Source AIContinue reading on Medium »

This seems to have gone under the radar: Accelerating Access to Research Results: New Implementation Date for the 2024 NIH Public Access Policy. It’s a memo from Jay Bhattacharya, director of the NIH (the United States’ National Institutes of Health): Well, this is tremendous news. The NIH is the biggest single funder of health research in the USA, and making all the work that it funds immediately open access is a huge win.

Rethinking How We Measure the Value of AI ResearchContinue reading on Medium »

We’ve been accelerating our metadata development efforts and recently released version 5.4 of our metadata schema, and are planning to release version 5.5 (including support for multiple contributor roles and the CRediT taxonomy) this summer. We will also extend our grants schema based on the Funders Advisory Group work, and make progress on other changes as set out on our new metadata development roadmap.

Following my recent talk for the Boston Library Consortium, many of you expressed a strong interest in learning how to test the new generation of AI-powered academic search tools. Specifically, evaluating systems using Retrieval-Augmented Generation (RAG) was the top request, surpassing interest in learning more about semantic search or LLMs alone. This is a crucial topic, as these tools are rapidly entering our landscape.

2024 was a banner year for dataset resources in DataCite. The National Institute of Fusion Science registered over 10,000,000 datasets. These metadata are remarkably complete - a bright spot demonstrating great metadata at scale.

In this dual case study, we learn why the Howard Hughes Medical Institute (HHMI) relies on OA.Report and why OA.Report relies on ROR to help HHMI track compliance with its open access policy. “Even back then [in 2019], the best option was to lean on a big, community-owned solution. And it’s been great to see ROR effectively become the standard, the clear way forward for identifying organizations.” “We think ROR is terrific.

Persistent Identification and Machine-Actionable KnowledgeContinue reading on Medium »

Seamless image compositing with SDXL and ComfyUIContinue reading on Medium »