Week 1 - Bioinformatics Focus - Gene-Disease Association Prediction

AIVIA Collaborations Team Workspace: Summer 2024
1

Project Scoping and Disease Selection

Objective: Define the project scope and select a single target disease for gene-disease association prediction.

Starting point: General project idea of predicting gene-disease associations using OpenTargets and STRING database data. Please check the Machine Learning Focus - Gene-Disease Association Prediction post to see detils of the initial data being used by the ML teams.

Tasks:

  • Select one specific disease for prediction:

    • Research disease categories in OpenTargets
    • Analyze data availability and quality for different diseases
    • Choose one disease based on criteria such as data abundance, research interest, and potential impact
    • Document selection criteria and rationale

  • Analyze data characteristics for the chosen disease:

    • Investigate the number of known gene associations
    • Assess the quality and completeness of available data
    • Identify any unique features or challenges associated with the selected disease

  • Define project scope:

    • Specify the exact prediction task (e.g., binary classification of gene-disease associations)
    • Outline the potential impact and applications of accurate predictions for the chosen disease

  • Create project documentation:

    • Write a detailed project objective
    • Justify the choice of the target disease
    • Specify evaluation criteria for the prediction task
    • Document relevant data sources (OpenTargets and STRING)

Expected outcome:

  • Selection of one target disease for prediction, with detailed justification
  • Analysis of data characteristics for the chosen disease
  • Clear project scope and objectives
  • Detailed project documentation including rationale for disease selection, objectives, and relevant data sources
1 Like
2

@Sanzida @Savann @moneuron @Thuraya_Ayman @hahaharsini @Moh_Saiger

Are you available to meet with Anya this weekend to discuss your findings? If you have a preference, please let us know below (pick all time slots that may work):

  • Sat (08/03) morning Pacific
  • Sat afternoon Pacific
  • Sat evening Pacific
  • Sun morning Pacific
  • Sun afternoon Pacific
  • Sun evening Pacific
0 voters

Please make sure to come prepared. We will use OpenTarget and STRINGdb for bulk data. OMIM does not permit scraping and requires a license for API, but we can get useful targeted data.

Meeting is an open one, all are welcome to attend.

3

Confirming Saturday 9am Pacific Meeting Time for the Bioinformatics Session

  • Please take a look at the tasks and come prepared for an interactive meeting.
  • If you are interested in this task but are not available for the Saturday meeting, please reply to this topic with project proposal and/or questions.
  • Participation in one of the above is mandatory to be part of the Bioinformatics subteam.

Zoom Link: Launch Meeting - Zoom

Check out the ML meeting times as well. The ML meeting for Sunday will be led by Sam and cover initial analysis of the bioinformics data.

4

Meeting Minutes (3 August 2024)

Project Definition IMG_52C8B67AABC5-1

Project Scope

  • Level 1: develop binary classification model focusing on only one disease (diabetes mellitus or breast cancer)
  • Level 2: integrate score/confidence level based on how much and type of evidence from OpenTargets and StringDB
  • Level 3: generalize model to work with multiple diseases

NOTE: you may need to explore StringDB’s API (Access - STRING functional protein association networks) to efficiently generate PPI networks for multiple diseases

PPI = protein-protein interaction

3 Likes