LinkMorph

Transform biomedical schemas to LinkML format

LinkMorph is a modular, extensible framework for converting various biomedical schema formats (FHIR, OMOP, graph schemas, JSON Schema, etc.) into LinkML format. It's designed specifically for biomedical data integration workflows.

Features

• Pluggable Architecture: Strategy pattern with processors for different schema types
• Biomedical-Aware: Built-in patterns for genomic, clinical, and research data
• Configurable: YAML-based configuration for domain-specific customization
• Graph Schema Support: Native support for vertex/edge graph schemas
• FHIR Ready: Handles FHIR resources and extensions
• Multi-Format: Supports JSON Schema, custom YAML schemas, and more
• Research-Friendly: Patterns for cohorts, studies, omics data, and clinical trials
• Schema-Centric Transformation Declarative field alignment using schema logic, not hardcoded scripts

Quick Start

Installation

git clone https://github.com/bmeg/linkmorph.git
cd linkmorph
pip install -e .

Basic Usage

linkmorph /path/to/iceberg/schemas/graph \
  --output-dir ./output \
  --project-name bmeg_iceberg \
  --verbose

linkmorph ./schemas \
  --config ./bmeg_config.yaml \
  --output-dir ./linkml_schemas

This generates:

• <project_name>_linkml_schema.yaml - main LinkML schema
• <project_name>_conversion_mapping.yaml - conversion documentation

Python API

import logging
logger = logging.getLogger(__name__)
from linkmorph import LinkMorphConverter, ConversionConfig

config = ConversionConfig.default_biomedical("my_project")
converter = LinkMorphConverter(config, logger)
output_path = converter.convert_directory("./input_schemas", "./output")

Supported Schema Types

Format	Description	Example Use Cases
Graph Schemas	Vertex/edge definitions	BMEG, knowledge graphs
FHIR Resources	HL7 FHIR format	Clinical data, EHRs
JSON Schema	Standard JSON Schema	APIs, data validation
Generic YAML	Custom YAML structures	Research data models

Configuration

Create a linkmorph_config.yaml file to customize behavior:

project_name: "my_biomedical_project"

type_mappings:
  gene_expression_value: "float"
  p_value: "float"

field_patterns:
  r'.*_cohort$': 'research_group'
  r'.*_pvalue$': 'statistical_significance'
  r'.*_chromosome$': 'genomic_location'

processing:
  auto_enum_threshold: 10
  ignore_patterns:
    - r'^_.*'