Navigating the Biomedical Literature: Insights vs. Papers

Target identification, a critical first step in drug development, is complicated by the complexity of diseases and the overwhelming amount of research being published. In this blog, we demonstrate how Causaly can manage this data overload to extract relevant scientific insights for Huntington’s disease (HD) from the entire volume of biomedical literature without bias.

By machine-reading the literature, Causaly instantly identified ~1,300 targets for HD. Focusing on the most novel insights, we prioritized targets reported in 2023-2024, uncovering almost 350 targets. Of these, around 85 have been explored in animal models. Further refinement based on primary data narrowed the search down to around 30 targets. In this example, we honed in on HD targets studied in mice, identifying hepatoma derived growth factor (HDGF) as a target with compelling evidence.

Taking these insights a step further, pathways which may mediate the effect of HDGF on HD were investigated using Causaly’s hypothesis generation tool (Figure 1). A study showed that by controlling HDGF, miR-939 has shown to deactivate the Wnt/β-catentin signal transduction pathway in prostate cancer.¹ As Wnt signaling has been implicated in HD pathogenesis,² it can be hypothesized that HDGF may mediate HD via this pathway.

We performed the same search in PubMed and compared it to the insights obtained using Causaly (Figure 2). PubMed returned a list of ~4,100 papers presented across 400+ pages, offering limited potential for efficient knowledge exploration. In comparison, Causaly found evidence for 1,300 targets extracted from 4,800+ documents, providing meaningful scientific insights rather than papers. Causaly leverages all scientific literature to return comprehensive results without bias, enabling scientists to uncover hidden connections and generate hypotheses.

With 2 publications added to PubMed every minute,³ identifying therapeutic targets with traditional keyword searching is time-consuming, causes reading fatigue and is subject to bias. By machine-reading the literature, Causaly manages this data overload, extracting insights rather than papers, to enable the exploration of more novel avenues.