Abstract B032: Generative modeling (AI) for the analysis of Single-Cell RNA-seq data of radioresistant malignant pediatric brain tumors
Bibliographic record
Abstract
Abstract Malignant pediatric brain tumors (MPBTs) are aggressive tumors that are the leading cause of mortality in pediatric neuro-oncology. Since surgical resection is often limited by tumor location, radiotherapy frequently serves as initial treatment. However, MPBTs might have spontaneously or can develop radioresistance (RR) where the cellular mechanisms underlying this resistance remain poorly understood. Single-Cell RNA sequencing (scRNA-seq), which characterizes tumor heterogeneity at the cellular level, might distinguish intrinsic and extrinsic specific transcriptomic signatures associated with RR before and after irradiation exposure. To investigate those relevant transcriptomic changes induced by irradiation, patient-derived tumoroids established from patient-derived xenograft models were exposed to a repetitive irradiation of 5x4Gys. scRNA-seq approach was performed on naïve tumoroids and 7 days after the end of irradiation. Based on this experimentation, we aim to identify gene expression changes occurring in response to irradiation. For this comparative purpose, we implemented a generative AI model based on Disentangled Variational Auto-Encoders (D-VAE) to learn irradiation-invariant representations of the transcriptome. These representations are then concatenated with controllable treatment attributes for data reconstruction. This model is generating realistic transcriptomic data of irradiated cells from both non-irradiated or irradiated tumor cells just by controlling the irradiation attribute sent to the generator of the model. The exploration of generated synthetic data through this model might be leveraged to understand the effects of irradiation treatments on gene expression profiles. From the generated transcriptomic data, our initial validation confirms that the D-VAE model accurately reproduces the radiation treatment effects and preserve cell clustering patterns. We are now exploring the opportunities provided by the ability to reconstruct "counterfactual" data to understand the fate of different cell clusters in response to irradiation. This process will allow to identify key genes and pathways that are consistently altered following irradiation or associated to RR. In particular, we are focusing on whether subset of cells display pre-existing RR biomarkers prior to treatment. Together, this work is demonstrating the utility of AI-driven approaches in modeling complex cellular responses to irradiation and is offering new insights of the molecular determinants of RR in MPBTs. Citation Format: Chinar Salmanli, Marlene Deschuyter, Natacha Entz-Werle, Julien Godet. Generative modeling (AI) for the analysis of Single-Cell RNA-seq data of radioresistant malignant pediatric brain tumors [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Artificial Intelligence and Machine Learning; 2025 Jul 10-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(13_Suppl):Abstract nr B032.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".