DOCSEARCHbyAlgolia
Submit Docs

GeneticScores.org Calculation Service

website·Indexed 32 days ago

https://docs.geneticscores.org

A service for calculating polygenic scores (PGS) sourced from the PGS Catalog. The platform provides tools for genome data upload via Globus Connect Personal or web browser, data encryption using crypt4gh (CLI and GUI), and a PGS Catalog Calculator including the pygscatalog tool.

Tokens
5.6K
Snippets
8
Records
52
Agent score
50%

What's inside GeneticScores.org

  1. Understand the GeneticScores.org public-key encryption workflow

    GeneticScores.org implements public-key cryptography via crypt4gh to secure datasets. The workflow is as follows:

    1. When a new dataset is created, a unique key pair (public key and secret key) is generated.
    2. The secret key is encrypted and stored in a secure vault on the GeneticScores.org platform.
    3. The public key is provided to the user for download. The public key is not sensitive.
    4. Data is encrypted using the public key; only the holder of the corresponding secret key (GeneticScores.org) can decrypt the data.

  2. Understand GeneticScores.org security and risk management practices

    GeneticScores.org implements a security strategy focused on identifying and mitigating risks related to confidentiality breaches, privacy/autonomy breaches, data corruption/destruction, and service disruption. Their approach includes:

    • Regular risk assessments and vulnerability scans.
    • Application penetration tests.
    • A defined protocol for breach response.
    • Adherence to the six GA4GH Control Objectives regarding unauthorized access, data integrity, service availability, and security attack prevention.

  3. Calculate polygenic scores using GeneticScores.org

    GeneticScores.org is a tool designed for AI developers and human health researchers to calculate polygenic scores from existing scoring files in the PGS Catalog. It applies weights from an existing polygenic score to variants present in user-supplied target genomes by running the PGS Catalog Calculator pipeline.

    Key Constraints:

    • It does NOT create new polygenic scores (variant/weight lists).
    • It does NOT estimate absolute risk.
    • It does NOT currently support genetic ancestry similarity analysis, although the underlying PGS Catalog Calculator does.

  4. Interpret the GeneticScores HTML report structure

    The HTML report provides a high-level overview consisting of three main sections:

    • Scoring file metadata: Information about the scores used (users should cite authors and check licenses).
    • Variant matching details: Describes the overlap between input target genomes and scoring files to ensure data quality.
    • Score distribution: A density plot summarizing the calculated scores.

  5. Use pygscatalog for PGS data processing

    pygscatalog is a collection of Python applications and libraries designed for working with polygenic scores and the PGS Catalog database. It provides the necessary metadata and utilities required for the accurate application and evaluation of polygenic scores. It is used internally by the pgsc_calc pipeline.
    # Source code available at: https://github.com/PGScatalog/pygscatalog

  6. Calculate polygenic scores using pgsc_calc

    The PGS Catalog Calculator (pgsc_calc) is a bioinformatics analysis pipeline used to calculate polygenic [risk] scores on samples with imputed genotypes. It supports the use of existing scoring files from the PGS Catalog as well as user-defined PGS/PRS. This tool is the underlying engine used by the GeneticScores.org calculation service when users upload target genomes.
    # Source code available at: https://github.com/PGScatalog/pgsc_calc
# Full documentation: https://pgsc-calc.readthedocs.io/en/v2.2.0/index.html

  7. Prepare genomic data for polygenic score calculation

    To calculate polygenic scores, upload a sample set of genomes or genotyping data. To avoid inflating accuracy metrics, ensure these samples are distinct from those used to develop the original polygenic scores (risk alleles and weights).

    Mandatory requirements for uploaded data:

    • Encryption: Must be encrypted using crypt4gh (public key available on the 'Transfer Your Data' page).
    • Format: Must be in PLINK1 (.bed, .bim, .fam) or PLINK2 (.pgen, .pvar, .psam) format.
    • Scope: Must contain chromosomes 1-22, X, and Y within a single fileset. Uploading data split by chromosome is not supported.
    • Reference Build: Must use human genome reference build GRCh38 or GRCh37 coordinates.

  8. Encrypt genetic data using crypt4gh-gui

    The crypt4gh GUI is used to encrypt data for the PGS calculation service. It is recommended for users who prefer not to use a terminal or are testing the platform. Note that the GUI can only encrypt one file at a time; for multi-file datasets (e.g., .psam, .pvar, and .pgen), each file must be encrypted individually. All data must be encrypted, including non-sensitive or public data, otherwise jobs will fail.

  9. Select polygenic scores by trait

    Users can select scores based on the traits they predict. GeneticScores.org uses 'mapped traits' (ontology terms from the Experimental Factor Ontology/EFO) to retrieve and group scores. Because 'reported traits' (author-provided descriptions) may be more specific than the mapped trait, users should inspect the reported trait in the PGS Catalog before applying scores.

  10. Workflow for using the GeneticScores.org Calculation Service

    To calculate genetic scores using the GeneticScores.org service, users must follow these sequential steps:

    1. Data Validation: Ensure target genomes meet the platform's data requirements and suggestions.
    2. Legal Compliance: Agree to the terms of use and sign a data processing agreement to grant consent for data processing.
    3. Data Preparation: Create a dataset and encrypt all files locally before uploading.
    4. Data Transfer: Upload the encrypted data using Globus.
    5. Score Selection: Choose the desired scores based on trait, publication, or a specific polygenic score ID.
    6. Retrieval: Download the final results report and the calculated scores.

  11. Handle multi-score pipeline submissions

    If a pipeline is submitted with more than one score, the pipeline is considered successful if at least one score runs successfully. In this case, you will receive results and a report. You must consult the report to determine which specific scores succeeded and which failed.

  12. Upload genome datasets via web browser

    Use the web browser upload method for smaller datasets (less than tens of gigabytes) or when you prefer not to install a dedicated transfer application. All transfers are protected via HTTPS encryption. For larger datasets, use the Globus Connect Personal application.

    Follow these steps to complete the upload:

    1. Enter your Globus account identity.
    2. Download the public encryption key and encrypt your data. Only encrypted data should be uploaded to ensure security.
    3. View the destination collection.
    4. Select the 'web upload' option.
    5. Grant the necessary permissions to the Globus web application.
    6. Select the files you wish to upload.
    7. Wait for the upload process to complete.
    8. Verify that all files have successfully arrived in the destination collection.