academia/F1000_talks/3/generation_task/instructions.md

You are to generate a complete, conference-quality academic slide deck suitable for an oral presentation at an academic conference, based strictly on the paper. The slides must be accurate, well-structured, and **faithful to the original paper**, with no fabricated content.

---

# **Strict Constraints for the Slides**

Below are the **hard constraints** you MUST satisfy. Slides violating these constraints are considered **incorrect**.

## 1. Content Requirements

The slide deck must have **16-20 slides**.

The slide deck must include the following sections, in the order listed below (the number of slides in each section may be determined as appropriate).

1. **Title Slide**

  * Paper Title: A multiscale functional map of somatic mutations in cancer integrating protein structure and network topology

  * Author Team: Yingying Zhang, Alden K. Leung, Jin Joo Kang, Yu Sun, Guanxi Wu, Le Li, Jiayang Sun, Lily Cheng, Tian Qiu, Junke Zhang, Shayne Wierbowski, Shagun Gupta, James Booth, Haiyuan Yu

  * Affiliation: Cornell University

  * Journal: bioRxiv (Preprint, 2024)

2. **Outline / Agenda**

3. **Introduction / Background**

  Cancer Mutational Analysis: Identifying functional driver mutations among millions of passenger mutations is the central challenge in cancer genomics.

  Current Landscape:
  Structural-based methods: Focus on localized mutation clustering in 3D protein structures.
  Network-based methods: Use protein-protein interaction (PPI) topology to find mutated gene modules.

  Motivation & Problem Statement: Existing methods are often non-overlapping and ignore the "edgetic" effect where mutations at specific interfaces affect only subset of interactions.

4. **Limitations of Existing Methods:**

  Structural Sparsity: Historically, only a small fraction of the human proteome had high-resolution experimental structures.

  Isotropic Propagation: Standard network models treat all interaction edges equally, failing to reflect that a mutation might disrupt one specific binding partner while leaving others intact.

  Low Signal-to-Noise: Gene-level mutation frequencies often fail to distinguish between functional clusters and random passenger mutations.

  Design Constraint: Include a visual example (refer to Fig 1b) showing how the "Human Protein Structurome" bridges the gap using AlphaFold and PIONEER.

5. **Overview of the Proposed Method**

  Core Idea: NetFlow3D—A unified framework that integrates atomic-level 3D mutation clustering with global network propagation.

  Key Contribution 1: Human Protein Structurome. A comprehensive repository covering 3D structures and binding interfaces for nearly all human proteins and PPIs.

  Key Contribution 2: 3D-Guided Initialization. Uses p-values from 3D mutation clusters to initialize "heat" in the network, boosting the signal of functional drivers.

  Key Contribution 3: Anisotropic Propagation. Weights network edges based on interface-specific structural evidence to model localized mutational impacts.

6. **Methodology: 3D Clustering & Structurome Construction**

  Step 1: Building the Structurome: Integrating AlphaFold2 models with PIONEER-predicted binding interfaces to map the "edgetic" landscape.

  Step 2: 3D Mutational Clustering: Identifying intra- and inter-protein clusters at the residue level, controlling for local background mutation rates.

7. **Key Algorithm: NetFlow3D Propagation**

  Heat Diffusion Model: Adapted from HotNet2, but enhanced with structural weights.

  Edge Anisotropy: Diffusion is directed more strongly towards interaction partners whose specific binding interfaces overlap with mutation clusters.

  Design Constraint: Display the construction process diagram (refer to Fig 1c-d) showing the flow from Atomic 3D Clustering -> Network Propagation -> Module Identification.

8. **Dataset and Technical Details**

  Data Sources: Somatic mutations from 9,946 tumors (33 cancer types) via TCGA MC3.

  Structural Scale: Covers 20,431 canonical protein structures and 146,316 PPI interfaces.

  Technical Stack: Leveraging high-performance computing for multiscale integration of atomic and topological data.

9. **Experimental Setup**

  Benchmarks: Evaluated against standard gene-level network analysis and 4 state-of-the-art 3D clustering algorithms.

  Validation: Functional testing via CRISPR-Cas9 fitness screens and TMT-IP-MS quantitative proteomics for specific modules.

10. **Experimental Results & Analysis**

  Increased Sensitivity: NetFlow3D identifies ~8-fold more proteins in significant modules compared to non-structural network methods.

  Biological Insight: Modules discovered are highly enriched for known cancer pathways and show strong associations with patient survival.

  Design Constraint: Include a performance comparison chart (refer to Fig 3c) demonstrating the superiority of 3D-integrated propagation over gene-level methods.

11. **Visual Analysis & Case Studies**

  Integrator-PP2A Complex: A novel module where interface mutations disrupt the recruitment of PP2A by the Integrator complex.

  Experimental Validation: Co-immunoprecipitation (Co-IP) confirming that PPP2R1A mutations (p.Arg258Cys) disrupt specific subunit interactions.

12. **Key Takeaways & Limitations**

  Takeaways: Multiscale integration is essential for mapping the functional consequences of mutations; "edgetic" disruptions are key drivers of cancer.

  Limitations: Performance relies on the accuracy of AlphaFold/PIONEER predictions; focus is limited to in-frame mutations.

13. **Conclusion**

  Summary: NetFlow3D provides a systematic map of how atomic-level mutations propagate through the cellular network to drive cancer.

  Future Work: Application to drug discovery by targeting specific disrupted interfaces identified by the framework.

---

## 2. Content Constraints

* **Faithfulness to background materials**: Use only the information in the paper. You must not fabricate additional experiments or modify or reinterpret the authors' claims.
* **Accuracy:** All content must be factually accurate, especially quantitative content and facts.
* **Brevity:** Use short, concise phrases, not long paragraphs. Focus on summarizing key facts and events without excessive detail. Bullet points may be used for clarity. If you use bullet points, each slide should have no more than 6 bullet points.
* **Sufficient Depth**: Do not summarize the paper in an overly superficial or high-level manner. The slides should preserve essential technical details, key arguments, and substantive insights rather than only presenting vague conclusions.
* **Logical Flow:** The slides should present a clear narrative, starting from early space exploration to recent developments. Ensure there is a clear progression of time and events.
* **Relevance of Information**: You must not add unrelated content.
* **Code & Markup Formatting**: Avoid raw LaTeX or Markdown code unless necessary.
* **Citation & Referencing**: Accurately reference the paper's results, diagrams, and examples. 
   * If a slide uses data from the paper, you must clearly indicate the source of the data on that slide (e.g., page xx, Figure xx, Table xx).
   * All references (if any) must be placed in the bottom-left corner of the slide.

## 3. Visual & Design

* **Images:** Include relevant images. Images must be high quality, clearly labeled, and relevant to the content.
* **Charts and Diagrams:** Use appropriate charts and diagrams where needed to visually present and clarify information, rather than relying only on text (and demos).
  * If the slide includes charts or figures, ensure that all visual elements are clearly annotated (e.g., axes are labeled, units are specified, legends are included where needed, and data points are explained when necessary).
  * Include **figures or diagrams descriptions** when appropriate, e.g., “The chart (from page 4 in the paper) shows proprietary models outperform open-weight ones.”
* **Legibility:** Use legible fonts and avoid clutter. Text should be large enough to be easily read.
* **Visual Balance:** Balance text and visuals so slides are easy to read when projected.
* **Layout:** Maintain a clean, professional layout with appropriate fonts, colors, and formatting.
* **Style Consistency**: The entire slide deck should follow a unified and coherent visual style.
* **Information Load**: Slides should avoid excessive information per page to preserve readability.

## 4. Text Quality

* All generated text should be clear, with no missing or incorrect characters or words.
* Spelling, grammar, and typography must be accurate and correct throughout the content.

## 5. Technical Fidelity Requirements

* **Quantitative Coverage**: Ensure that key data and experimental results (possibly presented in charts or tables in the paper) are included in the slide deck. In other words, the presentation should not only discuss the ideas of the paper but also present specific quantitative details (e.g., statistical data, experimental results, etc.).
   * The slide deck must include at least 5 slides with quantitative details.

* **Quantitative Detail Correctness**: Ensure quantitative details (task counts, benchmark size, etc.) are correct.

* **Table & Chart Traceability and Annotation**: Ensure that any figures and tables in your slide deck are consistent with the paper. Specifically, for every figure and table in the slides:
  * If it is directly copied from the paper, clearly indicate on the slide which figure or table it corresponds to in the paper (e.g., Figure 1 in the paper, Table 2 in the paper).
  * If it is newly plotted based on data from the paper, clearly specify which section of the paper the data are taken from (e.g., Section 3.1). In addition, provide a clear explanation of the meaning of each legend item in the figure and each row and column in the table.
  * For charts, every axis, unit, and label must be explicit

* **Point-Level Accuracy for Plots**: If scatter plots, line charts or radar charts are used in the slide deck, ensure that every data point exactly matches the corresponding data point in the original figure from the paper. Note that the values must be **precisely** the same, not just the shape of the graph.

* **Conceptual Illustration**: The slides may include data used only for conceptual illustration. However, if such data are included, you must clearly indicate on the corresponding slide which data are conceptual illustrations rather than experimental data reported in the paper.

## 5. Presentation Tone and Audience

* **Tone:** The tone should be informative, academic, and professional. It should avoid casual or informal conversational language, while remaining clear and suitable for oral presentation. The slide deck should maintain a consistent tone.
* **Audience:** The presentation is intended for an academic audience with relevant background knowledge in the field. The content should be accessible to graduate-level students and researchers, assuming familiarity with standard concepts and terminology, while still providing sufficient context to understand the motivation, methodology, and key contributions.

---

# **Output Expected**

A **complete slide deck** satisfying all constraints above.