Knowledge-Driven Mechanistic Enrichment of the Preeclampsia Ignorome

Tiffany J. Callahan, Adrianne L. Stefanski, Jin-Dong Kim, William A. Baumgartner Jr., Jordan M. Wyrwa, Lawrence E. HunterarXiv 2022 · OpenReview

Abstract

Preeclampsia remains a leading cause of maternal and fetal mortality, yet therapeutic progress is hindered by a significant "researcher bias" toward well-documented genes. This study utilizes a large-scale biomedical knowledge graph to identify and characterize the "preeclampsia ignorome"—a set of 445 genes experimentally associated with the disease but neglected in literature. By integrating transcriptomic data with heterogeneous biological networks, the researchers uncovered 53 novel, clinically relevant mechanistic associations, offering new avenues for drug repurposing and understanding disease pathogenesis.

Introduction

Preeclampsia is a severe pregnancy complication characterized by placental dysfunction, with delivery often being the only definitive treatment. Despite decades of high-throughput transcriptomic research identifying thousands of differentially expressed genes (DEGs), scientific focus remains concentrated on a small fraction of "popular" genes. This "researcher bias" means that the top 5% of genes account for 70% of published literature, leaving a vast "ignorome" of potentially critical biological drivers unstudied. This paper presents a systematic pipeline to prioritize these neglected genes by mapping them within a multi-domain biomedical knowledge graph (KG).

Methods

The researchers developed a multi-step computational pipeline to bridge the gap between experimental data and biological understanding:

  1. Ignorome Definition: A meta-analysis of human placental microarray data from the Gene Expression Omnibus (GEO) identified a molecular signature of 548 DEGs. Text-mining tools (PubTator, DisGeNET) were then used to filter out genes with existing preeclampsia associations. The remaining 445 genes (81% of the signature) were defined as the "ignorome."
  2. Knowledge Graph Construction: The study utilized PheKnowLator, a heterogeneous KG containing over 128,000 nodes and 3.2 million edges, representing relationships between genes, diseases, drugs, phenotypes, and pathways.
  3. Representation Learning: Node embeddings were generated using DeepWalk, converting the graph into a high-dimensional vector space where proximity indicates biological relatedness.
  4. Mechanistic Enrichment: For each ignorome gene, the 100 nearest neighbors across different domains were identified and subsequently validated by domain experts for biological plausibility.
Pipeline for identifying the preeclampsia ignorome and constructing the mechanistic knowledge graph.

Results

The analysis revealed a profound disconnect between experimental findings and follow-up research. While 548 genes were consistently altered in preeclampsia, over 80% had no prior literature link to the disease. Traditional enrichment tools often fail to characterize these genes because they rely on direct, documented links. In contrast, the KG approach successfully captured latent, multi-hop connections.

The study identified 53 novel mechanistic associations validated by molecular biologists, including:

  • 8 Disease Links: Identifying potential long-term health risks, such as a connection between the gene STS and ADHD in offspring.
  • 10 Drug Interactions: Suggesting repurposing opportunities, such as linking the gene MME to anti-asthmatic agents.
  • 10 Pathways & 5 Phenotypes: Mapping how placental dysfunction manifests metabolically. For instance, the gene PPM1K was linked to elevated branched-chain amino acids, a state known to impair mitochondrial respiration via mTORC1 signaling.
Literature coverage of preeclampsia ignorome genes across other disease categories, highlighting the potential for cross-disease insights.

Statistical validation confirmed these associations were highly significant ($p < 0.005$) compared to random samples.

Discussion and Future Work

This work demonstrates that biological networks can effectively prioritize neglected genes by uncovering hidden mechanistic links. However, the study is limited by its reliance on data up to 2017 and the need for manual expert review, which restricts scalability. Future iterations could leverage large language models (LLMs) to automate the formulation of mechanistic hypotheses. Ultimately, these computational predictions require "wet lab" validation to confirm their roles in preeclampsia pathogenesis and their potential as therapeutic targets.

Summary updated

Abstract

Preeclampsia is a leading cause of maternal and fetal morbidity and mortality. Currently, the only definitive treatment of preeclampsia is delivery of the placenta, which is central to the pathogenesis of the disease. Transcriptional profiling of human placenta from pregnancies complicated by preeclampsia has been extensively performed to identify differentially expressed genes (DEGs). The decisions to investigate DEGs experimentally are biased by many factors, causing many DEGs to remain uninvestigated. A set of DEGs which are associated with a disease experimentally, but which have no known association to the disease in the literature are known as the ignorome. Preeclampsia has an extensive body of scientific literature, a large pool of DEG data, and only one definitive treatment. Tools facilitating knowledge-based analyses, which are capable of combining disparate data from many sources in order to suggest underlying mechanisms of action, may be a valuable resource to support discovery and improve our understanding of this disease. In this work we demonstrate how a biomedical knowledge graph (KG) can be used to identify novel preeclampsia molecular mechanisms. Existing open source biomedical resources and publicly available high-throughput transcriptional profiling data were used to identify and annotate the function of currently uninvestigated preeclampsia-associated DEGs. Experimentally investigated genes associated with preeclampsia were identified from PubMed abstracts using text-mining methodologies. The relative complement of the text-mined- and meta-analysis-derived lists were identified as the uninvestigated preeclampsia-associated DEGs (n=445), i.e., the preeclampsia ignorome. Using the KG to investigate relevant DEGs revealed 53 novel clinically relevant and biologically actionable mechanistic associations.
BibTeX
@misc{callahan2022knowledge,
  author = {Tiffany J. Callahan and Adrianne L. Stefanski and Jin-Dong Kim and William A. Baumgartner Jr. and Jordan M. Wyrwa and Lawrence E. Hunter},
  title = {{Knowledge-Driven Mechanistic Enrichment of the Preeclampsia Ignorome}},
  year = {2022},
  eprint = {2207.14294},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2207.14294},
}
Figure