diff --git a/.github/actions/spelling/allow/names.txt b/.github/actions/spelling/allow/names.txt index 184503d0..73e97f2d 100644 --- a/.github/actions/spelling/allow/names.txt +++ b/.github/actions/spelling/allow/names.txt @@ -235,4 +235,5 @@ wlav zhang luciana melina -luque \ No newline at end of file +luque +Carlevaro \ No newline at end of file diff --git a/.github/actions/spelling/allow/terms.txt b/.github/actions/spelling/allow/terms.txt index c6bec148..0dfb8a5d 100644 --- a/.github/actions/spelling/allow/terms.txt +++ b/.github/actions/spelling/allow/terms.txt @@ -145,4 +145,10 @@ markdownify crcon crconlist petabyte -Vkv \ No newline at end of file +Vkv +apoptosis +cartopiax +Oncoprotein +oncoprotein +organoids +paraview \ No newline at end of file diff --git a/_data/contributors.yml b/_data/contributors.yml index ab8fee7c..4983279a 100644 --- a/_data/contributors.yml +++ b/_data/contributors.yml @@ -420,34 +420,33 @@ info: "Google Summer of Code 2025 Contributor" email: delatorregonzalezsalvador@gmail.com github: "https://github.com/salva24" - linkedin: "https://www.linkedin.com/in/salva-de-la-torre-gonzalez-7299b6335/" + linkedin: "https://www.linkedin.com/in/salvador-de-la-torre-gonzalez-7299b6335/" education: "Mathematics and Computer Engineering, University of Seville, Spain" active: 1 projects: - - title: "Agent-Based Simulation of CAR-T Cell Therapy Using BioDynaMo" - status: Ongoing + - title: "CARTopiaX: an Agent-Based Simulation of CAR T-Cell Therapy built on BioDynaMo" + status: Completed description: | - Chimeric Antigen Receptor T-cell (CAR-T) therapy has shown great promise - in treating hematological cancers by harnessing the immune system to target - and eliminate tumor cells. However, its effectiveness in solid tumors remains - limited due to challenges such as poor T-cell infiltration, immune suppression, - and T-cell exhaustion. This project aims to develop a scalable, agent-based - simulation of CAR-T therapy using BioDynaMo, a high-performance, open-source - biological simulation platform. The simulation will model key aspects of CAR-T - dynamics, including T-cell migration, tumor cell engagement, and the influence - of microenvironmental factors like hypoxia, regulatory T-cells, and - immunosuppressive cytokines. By simulating various tumor types, including solid - tumors and leukemias, the model will explore treatment variables such as CAR-T - dosage, timing of administration, and combination strategies like checkpoint - inhibitors and cytokine support. The project will deliver a complete and - well-documented BioDynaMo-based simulation modeling CAR-T cell therapy, - custom analysis scripts for visualizing tumor regression and evaluating CAR-T - efficacy, performance benchmarking across therapeutic strategies, and a detailed - report summarizing insights and future directions. This approach will provide - valuable insights for optimizing CAR-T therapies, particularly in complex tumor - microenvironments. + CAR T-cell therapy is a form of cancer immunotherapy that engineers a patient’s T cells + to recognize and eliminate malignant cells. Although highly effective in leukemias and other + hematological cancers, this therapy faces significant challenges in solid tumors due to the + complex and heterogeneous tumor microenvironment. CARTopiaX is an advanced agent-based model + developed to address this challenge, using the mathematical framework proposed in the Nature + paper “In silico study of heterogeneous tumour-derived organoid response to CAR T-cell therapy,” + successfully replicating its core results. Built on BioDynaMo, a high-performance, open-source + platform for large-scale and modular biological modeling, CARTopiaX enables detailed + exploration of complex biological interactions, hypothesis testing, and data-driven discovery + within solid tumor microenvironments. + The project achieved major milestones, including simulations that run more than twice as fast as + previous model, allowing rapid scenario exploration and robust hypothesis validation; high-quality, + well-structured, and maintainable C++ code developed following modern software engineering + principles; and a scalable, modular, and extensible architecture that fosters collaboration, + customization, and the continuous evolution of an open-source ecosystem. Altogether, this work + represents a meaningful advancement in computational biology, providing researchers with a powerful + tool to investigate CAR T-cell dynamics in solid tumor and accelerating scientific discovery while + reducing the time and cost associated with experimental wet-lab research. proposal: /assets/docs/de_la_torre_gonzalez_salvador_proposal_gsoc_2025.pdf - mentors: Vassil Vassilev, Lukas Breitwieser + mentors: Vassil Vassilev, Lukas Breitwieser, Luciana Melina Luque, Tobias Duswald - name: Rohan Timmaraju photo: Rohan_Timmaraju.jpg diff --git a/_data/crconlist2025.yml b/_data/crconlist2025.yml index a0812ec0..390dfb23 100644 --- a/_data/crconlist2025.yml +++ b/_data/crconlist2025.yml @@ -92,7 +92,7 @@ connect: "[Link to zoom](https://princeton.zoom.us/j/97915651167?pwd=MXJ1T2lhc3Z5QWlYbUFnMTZYQlNRdz09)" label: crcon25_part_1 agenda: - - title: "CARTopiaX an Agent-Based Simulation of CAR -T -Cell Therapy built on BioDynaMo" + - title: "CARTopiaX an Agent-Based Simulation of CAR T-Cell Therapy built on BioDynaMo" speaker: name: "Salvador de la Torre Gonzalez" time_cest: "15:00 - 15:20" @@ -121,7 +121,7 @@ tumors and accelerating scientific discovery while reducing the time and cost associated with experimental wet-lab research. - # slides: /assets/presentations/... + slides: /assets/presentations/Salva_GSoC25_final_presentation_CART.pdf - title: "Efficient LLM Training in C++ via Compiler-Level Autodiff with Clad" speaker: diff --git a/_data/standing_meetings.yml b/_data/standing_meetings.yml index cb616a5f..624a30f9 100644 --- a/_data/standing_meetings.yml +++ b/_data/standing_meetings.yml @@ -425,3 +425,7 @@ date: 2025-08-14 15:00:00 +0200 speaker: "Salvador de la Torre Gonzalez" link: "[Slides](/assets/presentations/Salva_GSoC_midterm_presentation_CART.pdf)" + - title: "CARTopiaX: an Agent-Based Simulation of CAR T-Cell Therapy built on BioDynaMo. GSoC Wrap-Up: CompilerResearchCon 2025" + date: 2025-10-30 15:00:00 +0200 + speaker: "Salvador de la Torre Gonzalez" + link: "[Slides](/assets/presentations/Salva_GSoC25_final_presentation_CART.pdf)" diff --git a/_posts/2025-10-31-CARTopiaX-ABM-CART-therapy-BioDynaMo.md b/_posts/2025-10-31-CARTopiaX-ABM-CART-therapy-BioDynaMo.md new file mode 100644 index 00000000..1f005169 --- /dev/null +++ b/_posts/2025-10-31-CARTopiaX-ABM-CART-therapy-BioDynaMo.md @@ -0,0 +1,129 @@ +--- +title: "CARTopiaX: an Agent-Based Simulation of CAR T-Cell Therapy built on BioDynaMo" +layout: post +excerpt: "CARTopiaX is an agent-based model developed using BioDynaMo to simulate CAR-T cell dynamics and interactions within the solid tumor microenvironment. It provides researchers with a tool to efficiently evaluate therapy efficacy and identify strategies to improve treatment outcomes." +sitemap: true +author: Salvador de la Torre Gonzalez +permalink: blogs/gsoc25_salvador_wrapup_blog/ +banner_image: /images/blog/cartopiax_plasma_gsoc.png +date: 2025-10-31 +tags: gsoc BioDynaMo c++ +--- + +## Project Overview + +**CAR T-cell therapy** is a form of cancer immunotherapy that engineers a patient’s T cells to recognize and destroy malignant cells. While this approach has achieved remarkable success in treating blood cancers, it faces significant challenges in solid tumors due to the complexity and heterogeneity of their microenvironments. + +**CARTopiaX** is an **agent-based model** designed to simulate the behavior of *tumour-derived organoids* which are lab-grown models that mimic real solid tumor environments and their response to CAR T-cell therapy. The project aims to bridge the gap between laboratory experiments and computational biology by developing a high-fidelity **in silico digital twin** of tumor-derived organoids for studying solid tumor dynamics and immunotherapy outcomes.
+Built on **BioDynaMo**, a high-performance, open-source platform for large-scale biological modeling, CARTopiaX enables researchers to: + +- Recreate realistic *in vitro* conditions for tumor growth. +- Introduce CAR T-cells and evaluate their efficacy in heterogeneous, solid tumor microenvironments. +- Explore different therapeutic strategies and parameter variations. +- Assess treatment outcomes such as tumor reduction, elimination, or relapse risk. + +CARTopiaX implements the mathematical framework described in the *Nature* publication *“In silico study of heterogeneous tumour-derived organoid response to CAR T-cell therapy”,* successfully replicating its key results and extending them through improved scalability and performance. + +Project Highlights: + +- **Performance:** Simulations run more than twice as fast as previous existing model, enabling rapid scenario exploration and hypothesis validation. +- **Software Quality:** Developed in **C++** following robust software engineering practices, ensuring high-quality, maintainable, and efficient code. +- **Architecture:** Designed to be scalable, modular, and extensible, fostering collaboration, customization, and continuous evolution within an open-source ecosystem. + +--- +## Model Replication + +All plots in this section compare CARTopiaX with the model results from the *Nature* paper over five runs, demonstrating **successful replication** and reproducing the same **biological findings**. Even though the graphs do not always overlap, this is due to substantial known differences in their modeling approaches and stochastic nature. What is important is that the overall behaviors and key biological dynamics are accurately reproduced, as researchers primarily focus on these trends and peak responses when designing treatments. + +### Replication Example: Tumor with no CAR-T treatment + +The following example reproduces the 30-day evolution of a 150 µm radius tumor simulation with no CAR T-cell treatment applied. The plot below shows the total number of cancer cells and the tumor radius over time: + +No treatment cells and radius + +Tumor Heterogeneity: + +CARTopiaX models a **heterogeneous tumor**, where cancer cells differ in their *oncoprotein expression levels*, representing varying proliferative capacities. +Although oncoprotein levels are continuous, cells are grouped into four discrete categories: + +| Type | Oncoprotein Level | Aggressiveness | +|:----:|:---------------------:|:--------------------:| +| 1 | 1.5 < Oncoprotein ≤ 2 | Most proliferative | +| 2 | 1 < Oncoprotein ≤ 1.5 | Very proliferative | +| 3 | 0.5 < Oncoprotein ≤ 1 | Less proliferative | +| 4 | 0 ≤ Oncoprotein ≤ 0.5 | Least proliferative | + +Tumor evolution visualized in ParaView: + +No treatment ParaView + +Type 1 cells are highly proliferative, causing their proportion in the tumor to increase, while Type 3–4 cells divide less frequently and, struggling in a oxygen-limited, resource-competitive environment, gradually become less common: + +No treatment Type of Cells + +As cell quantity rises, **oxygen levels drop** due to increased consumption, while the average **oncoprotein level gradually increases** because more proliferative cells pass their higher oncoprotein levels to their progeny over time. + +No treatment Oxygen and Oncoprotein + + +### Replication Example: One CAR T-cell dose of Scale 1:1 + +The following example models the same 150 µm-radius tumor, but with a single CAR T-cell dose, equal in number to the tumor cells administered, on day 0. The CAR T-cell population then declines stochastically due to apoptosis, reaching minimal levels around day 10. The plot below shows the total tumor and CAR T-cell populations over time: + +1Dose Tumor and CAR T-cell amount + +All cells consume oxygen, and cancer cells can die from **necrosis** in its absence. In addition, CAR T-cells tend to **eliminate higher oncoprotein-expressing cells more effectively**, leading to the following dynamics: + +Before day 10: CAR T-cells are actively killing tumor cells and **Type 1 and 2 cell populations decrease** as more proliferative cancer cells are targeted. + +After day 10: When CAR T-cells die from apoptosis, **Type 1 and 2 cells increase** their proportion in the tumor at the expense of Type 3 and 4 cells, as high-oncoprotein-expressing cells divide faster. + +1dose Percentage of Each Cell Type + +Before day 10: In the beginning, **oxygen** levels decrease with CAR T-cell arrival. Then they **gradually rise** as both CAR T-cells and tumor cells die, reducing overall consumption. On the other hand, the **average oncoprotein level drops** rapidly due to CAR T-cell preferential elimination of the most aggressive cancer cells. + +After day 10: CAR T-cells are no longer present and therefore tumor resumes growth, making **oxygen levels decline**. In addition, **oncoprotein levels rise** as highly proliferative cells dominate once again. + +1dose Oncoprotein and Oxygen Levels + +During the simulation, **dead and resistant cells** accumulate around the tumor core, forming a ***shield-like barrier*** that impedes CAR T-cell infiltration and reduces treatment effectiveness. + +Cell Shield Formation + +A 3D visualization of the sliced tumor and CAR T-cell treatment used in this example, rendered in *ParaView*, is available here: +[Watch the simulation video](https://youtu.be/7V8n627Nmzc) + +### Replication Example: Less is better, increasing cellular dosage does not always increase efficacy + +We consider again the 150 µm-radius tumor and now **two CAR T-cell doses of scale 1:1** (equal in number to the initial tumor cells) are delivered on days 0 and 8. This way much less tumor cells are present at day 30 in comparison to applying just one dose: + +2Doses of Scale 1:1 + +Delivering more CAR T-cells has helped in controlling the tumor, however this is not always the case. The next plot shows the same initial tumor but with two CAR T-cell doses containing a quantity twice the initial tumor cell count, on days 0 and 5. By day 30, the number of **tumor cells is roughly the same** as before, despite using **twice the amount of CAR T-cells**. + +2Doses of Scale 2:1 + + +Increasing CAR T-cell dosage does not necessarily improve tumor killing and can increase *toxicity*. The model suggests two doses at a 1:1 CAR T-to-cancer cell ratio, balancing effectiveness and safety, minimizing inactive *free* CAR T-cells. + +### Performance + +CARTopiaX successfully replicates the findings described in the *Nature* publication, achieving a **2× speed improvement** compared to the previous implementation. +This performance gain enables faster scenario exploration and larger-scale simulations. + +Execution Time Comparison + +--- + +## Conclusion + +By building a BioDynaMo-based model of CAR T-cell therapy, we provide a flexible, high-performance tool for exploring treatment strategies in complex tumor environments. CARTopiaX combines **speed**, **robust software quality**, and a **modular, open-source architecture**, making it highly **extensible** and **adaptable** for future research. Altogether, it offers the scientific community an efficient framework that accelerates discovery, guides experimental design, and supports the development of more effective immunotherapies for solid tumors. + +--- + +## Related Links + +- [Project Repository](https://github.com/compiler-research/CARTopiaX) +- [BioDynaMo Repository](https://github.com/BioDynaMo/biodynamo) +- Luque, L.M., Carlevaro, C.M., Rodriguez-Lomba, E. et al. *In silico study of heterogeneous tumour-derived organoid response to CAR T-cell therapy.* Sci Rep 14, 12307 (2024). [DOI](https://doi.org/10.1038/s41598-024-63125-5) +- [My GitHub Profile](https://github.com/salva24) diff --git a/assets/presentations/Salva_GSoC25_final_presentation_CART.pdf b/assets/presentations/Salva_GSoC25_final_presentation_CART.pdf new file mode 100644 index 00000000..022a491c Binary files /dev/null and b/assets/presentations/Salva_GSoC25_final_presentation_CART.pdf differ diff --git a/images/blog/cartopiax/dose_scale1_day0_dose_scale1_day8_num_cells.png b/images/blog/cartopiax/dose_scale1_day0_dose_scale1_day8_num_cells.png new file mode 100644 index 00000000..ee8afa43 Binary files /dev/null and b/images/blog/cartopiax/dose_scale1_day0_dose_scale1_day8_num_cells.png differ diff --git a/images/blog/cartopiax/dose_scale1_day0_num_cells.png b/images/blog/cartopiax/dose_scale1_day0_num_cells.png new file mode 100644 index 00000000..d6ddf6d0 Binary files /dev/null and b/images/blog/cartopiax/dose_scale1_day0_num_cells.png differ diff --git a/images/blog/cartopiax/dose_scale1_day0_oncoprotein_and_oxygen.png b/images/blog/cartopiax/dose_scale1_day0_oncoprotein_and_oxygen.png new file mode 100644 index 00000000..7fcf4189 Binary files /dev/null and b/images/blog/cartopiax/dose_scale1_day0_oncoprotein_and_oxygen.png differ diff --git a/images/blog/cartopiax/dose_scale1_day0_type_of_cells_absolute_numbers.png b/images/blog/cartopiax/dose_scale1_day0_type_of_cells_absolute_numbers.png new file mode 100644 index 00000000..ad4e7167 Binary files /dev/null and b/images/blog/cartopiax/dose_scale1_day0_type_of_cells_absolute_numbers.png differ diff --git a/images/blog/cartopiax/dose_scale1_day0_type_of_cells_percentage_populations.png b/images/blog/cartopiax/dose_scale1_day0_type_of_cells_percentage_populations.png new file mode 100644 index 00000000..3ad4c46e Binary files /dev/null and b/images/blog/cartopiax/dose_scale1_day0_type_of_cells_percentage_populations.png differ diff --git a/images/blog/cartopiax/dose_scale2_day0_dose_scale2_day5_num_cells.png b/images/blog/cartopiax/dose_scale2_day0_dose_scale2_day5_num_cells.png new file mode 100644 index 00000000..d47d1570 Binary files /dev/null and b/images/blog/cartopiax/dose_scale2_day0_dose_scale2_day5_num_cells.png differ diff --git a/images/blog/cartopiax/execution_times.png b/images/blog/cartopiax/execution_times.png new file mode 100644 index 00000000..52e0b802 Binary files /dev/null and b/images/blog/cartopiax/execution_times.png differ diff --git a/images/blog/cartopiax/model_outline.png b/images/blog/cartopiax/model_outline.png new file mode 100644 index 00000000..ae7a43e4 Binary files /dev/null and b/images/blog/cartopiax/model_outline.png differ diff --git a/images/blog/cartopiax/no_dose_tumor_evolution_paraview.png b/images/blog/cartopiax/no_dose_tumor_evolution_paraview.png new file mode 100644 index 00000000..cbe298be Binary files /dev/null and b/images/blog/cartopiax/no_dose_tumor_evolution_paraview.png differ diff --git a/images/blog/cartopiax/no_doses_num_cells_and_tumor_radius.png b/images/blog/cartopiax/no_doses_num_cells_and_tumor_radius.png new file mode 100644 index 00000000..a0663fc7 Binary files /dev/null and b/images/blog/cartopiax/no_doses_num_cells_and_tumor_radius.png differ diff --git a/images/blog/cartopiax/no_doses_oncoprotein_and_oxygen.png b/images/blog/cartopiax/no_doses_oncoprotein_and_oxygen.png new file mode 100644 index 00000000..08365322 Binary files /dev/null and b/images/blog/cartopiax/no_doses_oncoprotein_and_oxygen.png differ diff --git a/images/blog/cartopiax/no_doses_type_of_cells_absolute_numbers.png b/images/blog/cartopiax/no_doses_type_of_cells_absolute_numbers.png new file mode 100644 index 00000000..1289883c Binary files /dev/null and b/images/blog/cartopiax/no_doses_type_of_cells_absolute_numbers.png differ diff --git a/images/blog/cartopiax/shield.png b/images/blog/cartopiax/shield.png new file mode 100644 index 00000000..458055f0 Binary files /dev/null and b/images/blog/cartopiax/shield.png differ diff --git a/images/blog/cartopiax_plasma_gsoc.png b/images/blog/cartopiax_plasma_gsoc.png new file mode 100644 index 00000000..d94d1625 Binary files /dev/null and b/images/blog/cartopiax_plasma_gsoc.png differ