💊

Demo of Jules OS Drug Analysis

Aug 30, 2024

Notes on Jules OS Demo

Overview

  • Presenter: Adrian
  • Focus: Demonstration of Jules OS answering a query about the top-selling drugs.

Initial Query

  • User asks for revenue of the 10 top-selling drugs targeting proteins highly expressed in the liver.
  • Jules OS begins processing the query immediately.

Plan Formulation

  • Jules OS outlines a plan to tackle the question.
  • A mini-map shows the sequential steps involved in the process.
  • User prompted to proceed to the first step.

Step 1: Retrieve Liver-Expressed Proteins

  • Jules OS executes the first step to gather proteins highly expressed in the liver.
  • Actions Taken:
    • Retrieves genes dataset.
    • Designs SQL query for data extraction.
    • Filters for liver-expressed proteins.
  • A preview of the extracted data is shown for evaluation.
  • User approves and moves to the next step.

Step 2: Drug-Protein Interactions

  • Next step involves obtaining drug data linked to the proteins.
  • Actions Taken:
    • Retrieves drug data.
    • Designs and executes SQL query for extraction.
  • Data extraction is successful; user proceeds.

Step 3: Sales Information

  • Final data extraction step focuses on sales data for all drugs.
  • User checks minimap; confirms successful extraction.

Data Analysis and Processing

  • Jules OS analyzes the collected data to answer the original question.
  • Key Actions:
    • Writes and executes Python code to process the data.
    • Breaks down data wrangling into sub-steps.

Sub-Step 1: Merging Gene and Drug Data

  • Merges the extracted gene data with drug data.
  • Sample of output data is reviewed; looks good.

Sub-Step 2: Merging Sales Data

  • Attempts to merge extracted sales data; initial attempt yields an empty data frame.
  • Troubleshooting:
    • Adjusts naming conventions (normalizes names by converting to lowercase and removing spaces).

Sub-Step 3: Sorting Results

  • Final sorting step to display top-selling drugs by sales in 2022.
  • Updates plan to include target protein names alongside drug sales data.

Conclusion

  • Jules OS generates a list of the top 10 selling drugs targeting liver-expressed proteins.
  • Emphasizes the capability of Jules OS to provide complex data insights without requiring coding knowledge from the user.
  • Highlights the potential to enhance research productivity at GSK.