Lacking mechanistic disease definitions and corresponding association data hamper progress in network medicine and...

Area of Science:

• Network Medicine
• Computational Biology
• Systems Biology
• Genomics

Background:

• Network medicine seeks to replace traditional phenotype-based disease classifications with subtypes linked to specific pathomechanisms.
• Current approaches often utilize large-scale disease association data, which may be annotated using the very phenotype-based definitions network medicine aims to transcend.
• This reliance on potentially inadequate annotations raises concerns about biases in pathomechanism discovery.

Purpose of the Study:

• To investigate the extent to which biases from phenotype-based disease annotations distort pathomechanism mining in network medicine studies.
• To evaluate the impact of current disease annotation practices on the reliability of network-derived pathomechanisms.

Main Methods:

• Construction of networks using diverse types of disease association data.
• Global- and local-scale analyses of these constructed networks.
• Comparative analysis of network mining results based on different data annotation strategies.

Main Results:

• Results indicate that large-scale disease association data, when annotated with phenotype-based definitions, can introduce significant biases.
• These biases can distort the identification of distinct pathomechanisms, potentially leading to mechanistically inadequate conclusions.
• The study highlights the limitations of relying solely on broad disease association data for uncovering precise disease mechanisms.

Conclusions:

• Large-scale disease association data should be utilized with caution for pathomechanism mining in network medicine.
• Analyses of such data necessitate complementary close-up investigations using molecular data from well-characterized patient cohorts.
• Integrating molecular insights is crucial for validating and refining network-derived disease mechanisms and overcoming annotation biases.