Machine Learning and Drug Discovery at @AstraZeneca (views my own). PhD @durham_uni, passed through @computingncl, @gliffai, @CriteoAILab
Joined May 2014
- Tweets286
- Following361
- Followers173
- Likes156
Stephen Bonner retweeted
Broadcasting, reshaping, repeating, and squeezing in @PyTorch is tricky business. Thanks to @SBonner0 for identifying some room for improvement in our implementation of Margin Ranking loss
🔗 PR: github.com/pykeen/pykeen/pul…
#machinelearning #ai #artificialintelligence #opensource
Stephen Bonner retweeted
🚨 PyKEEN 1.7.0 is out! 🚨
New: 7 models, 3 encoders, 2 trackers, keras-style training callbacks, uncertainty estimation, and more
📖 Full Release Notes: github.com/pykeen/pykeen/rel…
💪 Huge S/O to external contributors @dobraczka @michael_galkin @SBonner0 @Ralph_Abb and others
Happy to present a new paper today at @lmrl_bio, held at @NeurIPSConf. The paper explores how topological imbalance in drug discovery knowledge graphs can affect models trained upon them and is joint work with Ufuk Kirik, Ian Barrett, Ola Engkvist and Jian Tang.
Stephen Bonner retweeted
Delighted to have won the Best Partnership Alliance for our work with @AstraZeneca at the prestigious #ScripAwards.
Our teams work side-by-side to combine AI & scientific expertise to discover novel targets for #CKD & #IPF.
Learn more: benevolent.com/news/a-new-er… @PharmaScrip
An updated version of "A Review of Biomedical Datasets Relating to Drug Discovery: A Knowledge Graph Perspective" with I Barrett, C Ye, R Swiers, O Engkvist, @AndreasBenderUK, @cthoyt and W Hamilton has been released.
Paper: shorturl.at/ehDS0
Repo: shorturl.at/fgC24
New for this version is the associated github repo collecting all the relevant resources, papers and dataset links into one location. This repo will continue to be updated as new resources are released.
For those having issues with the shortened paper URL, here in the full link: arxiv.org/abs/2102.10062
Stephen Bonner retweeted
Want to learn about applications of graphs in the industry?
🚀Then, register for our Graph #MachineLearning in Industry workshop on 23rd September⬇️
sites.google.com/view/graph-…
Stephen Bonner retweeted
Global disparities in COVID-19 genomic surveillance between high-, middle- and low-income countries 🦠🌎🧬. See original post by @AndersonBrito_ and the preprint medrxiv.org/content/10.1101/…. Please share your thoughts, comments and questions with us!
Stephen Bonner retweeted
EMBL and @DeepMind have partnered – a breakthrough for science.
Together, we're providing a treasure trove of protein structure predictions powered by #AlphaFold to herald a new era for #AI-enabled biology.
embl.org/news/science/alphaf…
ALT Protein structures representing the data obtained via AlphaFold.
Stephen Bonner retweeted
We are happy to see that PyKEEN is adopted within research and industry projects. Recently, PyKEEN has been used within @AstraZeneca to analyze the potential of knowledge graph embedding models for drug discovery: arxiv.org/pdf/2105.10488.pdf
#knowledgegraphembeddings
Stephen Bonner retweeted
Was reading the @benevolent_ai knowledge graph paper by @sgpaliwal et al and thought it would help my understanding to draw a metagraph. So here it is:
Paper: doi.org/gg49
Metagraph: git.io/JZOHB
ALT Metagraph of the Benevolant AI knowledge graph in Paliwal et al 2020
Stephen Bonner retweeted
We just merged a nice external contribution from @SBonner0 to use @TensorBoard as a tracker 🚀
📖 Tutorial: pykeen.readthedocs.io/en/lat…
🖥️ Code: github.com/pykeen/pykeen/pul…
Stephen Bonner retweeted
Researchers at @AstraZeneca's Data Sciences & Quantitative Biology Unit combine #Hetionet, #Pharmaprojects, and @OpenTargets to predict disease-target relationships. Excited to give this a deeper read! arxiv.org/abs/2105.10578
ALT Figure 3: Schema of the proposed model.