#AACRprecmed25

I just realized I had #AACRprecmed25 stuck on my live-posting. None of this has anything to do with that conference! Oops

JB: Cells of the early fracture gap. scRNA-seq #AACRprecmed25 www.biorxiv.org/content/10.1101/2025.01....

JB: Cells of early fracture site are not hypoxic. Vascular transport is not the oxygen source. #AACRprecmed25

JB: Test with a euthanized control animal. Looks like EF5 is getting to the fracture site but the fracture site is high in O₂ #AACRprecmed25

Continued 1: JB: EF5 marks hypoxic cells in vivo. Why doesn't it mark cells in a hematoma? #AACRprecmed25 🧪🧬🖥️

JB shows data demonstrating EF5 marking low O₂ in vitro #AACRprecmed25

JB: EF5 is a nitroimidazole that marks hypoxic cells. Under hypoxia it is reduced to adducts that can be identified with an antibody. #AACRprecmed25

JB: Which cells in the fracture gap are hypoxic? How does this hypoxia affect their mechanobiology? #AACRprecmed25

JB: Endochondral regeneration. Early loading disrupts vascular ingrowth. Blood vessels cannot grow in. Early loading promotes cartilage formation. #AACRprecmed25 www.science.org/doi/10.1126/scitranslmed...

JB: In vivo mechanical loading. Early loading disrupted bone formation. Delayed loading led to significant increase in bone formation. #AACRprecmed25

JB: How does hypoxia influence response to mechanical cues? #AACRprecmed25 https://pmc.ncbi.nlm.nih.gov/articles/PMC4464690/

JB: Physiologic oxygen tension. Ambient air is 21% O₂. But that's not "normoxic" for ourselves. Brain is about 4% O₂, up to arterial blood 13% O₂. Spleen is 8%. #AACRprecmed25

JB: Hypoxia promotes chondrogenesis in vitro through HIF signaling. #AACRprecmed25

Fracture healing under healthy and inflammatory conditions - Nature Reviews Rheumatology The interplay between the cells that regulate bone architecture and the immune system is increasingly recognized. In this Review, as well as providing an overview of fracture treatment and healing, the authors discuss our current knowledge of the part played by inflammation in the fracture repair process. The influence of biomechanical and biological factors on bone healing is also considered, focusing on the effects of excessive local and systemic inflammation, as occurs in autoimmune diseases such as rheumatoid arthritis.

JB: "Fracture leads to blood vessel rupture… and haematoma… This haematoma is characterized by hypoxia and low pH." #AACRprecmed25

JB discussing vascularity, oxygen, and fracture repair #AACRprecmed25 https://www.pnas.org/doi/10.1073/pnas.1107019108

JB: Algebra, like all good ideas, came from orthopedics. Muhammad al-Khwarizmi's Al-jabr, the setting of bones by compression, around 820. #AACRprecmed25

Things are tough all around, don't get me wrong. Yet, I just spent the better part of a week in Boston focused on precision medicine #AACRprecmed25 and feel so recharged. The ongoing research holds such incredible promise for both rare and common cancers. If only the grip on the NIH would loosen..

This afternoon, #AACRprecmed25 cochairs Anthony G. Letai, Elaine R. Mardis, Peter Horak, and Alice Soragni adjourned the AACR Special Conference on Functional and Genomic Precision Medicine in Cancer. We thank the chairs for developing an outstanding program.

It was a pleasure to attend #AACRprecmed25. Have to catch a plane now. Thanks to @theaacr.bsky.social @alice.soragnilab.com and the other organizers. Great meeting!

EvA: Q: Best drug for patient in front of you. EvA: Need biomarkers *and* drugs. And biology. Some sort of single-cell or spatial assay and a functional readout that points to downstream properties #AACRprecmed25

EvA: Q: If you were building a precision medicine endeavor from scratch, where would you allocate your dollar? EvA: Depends on the task. #AACRprecmed25

The Molecular Oncology Almanac

Continued 2: EvA: Molecular oncology almanac is available online [this is cool] #AACRprecmed25 🧪🧬🖥️

EvA: A molecular oncology almanac #AACRprecmed25 www.nature.com/articles/s43...

EvA: Multi-modal inference data for improved patient-specific predictions. #AACRprecmed25 www.cell.com/cell-reports-medicine/fu...

Deep learning-based predictions of gene perturbation effects do not yet outperform simple linear baselines Advanced deep-learning methods, such as foundation models, promise to learn representations of biology that can be employed to predict in silico the outcome of unseen experiments, such as the effect of genetic perturbations on the transcriptomes of human cells. To see whether current models already reach this goal, we benchmarked five foundation models and two other deep learning models against deliberately simplistic linear baselines. For combinatorial perturbations of two genes for which only the individual single perturbations had been seen, we find that the deep learning-based approaches did not perform better than a simple additive model. For perturbations of genes that had not yet been seen, the deep learning-based approaches did not outper-form the baseline of predicting the mean across the training perturbations. We hypothesize that the poor performance is partially because the pre-training data is observational; we show that a simple linear model reliably outperforms all other models when pre-trained on another perturbation dataset. While the promise of deep neural networks for the representation of biological systems and prediction of experimental outcomes is plausible, our work highlights the need for clear setting of objectives and for critical benchmarking to direct research efforts. Contact constantin.ahlmann{at}embl.de ### Competing Interest Statement The authors have declared no competing interest.

EvA: Caution: deep learning methods might not be better than simple models. Paper by @s-anders.bsky.social @wkhuber.bsky.social #AACRprecmed25

EvA: General uncertainty on if or how to use single-cell foundation models for cancer purposes. May be helpful for annotation. Not sure if fancy AI is the answer yet. Less clear of utility for fine-tuning on harder tasks. #AACRprecmed25

EvA: Integrating functional biology knowledge and ML for molecular discovery. Models: standard methods (HVG, scVI, PCA), foundation models (Geneformer, scGPT, scPrint) #AACRprecmed25

EvA: Your favorite signaling pathway is incorrectly annotated in Reactome. [see how incorrect it is outside Reactome!] #AACRprecmed25

EvA: Engineered known biology into the network. Good performance in precision-recall analysis. #AACRprecmed25

Biologically informed deep neural network for prostate cancer discovery - Nature A biologically informed, interpretable deep learning model has been developed to evaluate molecular drivers of resistance to cancer treatment, predict clinical outcomes and guide hypotheses on disease progression.

EvA: P-NET: sparse neural network with engineered architecture encoding biological processes. #AACRprecmed25