Exploring Biochemical Pathways with Human-Centric AI

In drug development, understanding biochemical pathways is essential for elucidating the molecular mechanisms underlying cellular and tissues functions, thereby informing targeted therapeutic strategies. In this example, we used Causaly to identify and explore biochemical pathways affected by targets for acute kidney injury (AKI).

Over 2,000 targets for AKI, extracted from 9,700+ papers, were identified using Causaly and can be visualized as a dendrogram view of results, as shown in Figure 1.

Targets were prioritized by those studied in animal models in the last 5 years, uncovering almost 1,000 targets. Pigs are attractive models for AKI in humans due to their large size and shared physiological patterns with humans,¹ targets were further prioritized by those studied in porcine models, revealing almost 40 targets, the majority of which were reported in primary data. By examining the evidence presented by Causaly, NOX2 (also known as CYBB) and NOX4 were selected as potentially interesting targets for further exploration.

NOX2 and NOX4 enzymes are major sources of reactive oxygen species (ROS) in the kidney and their upregulation has been implicated in both AKI and CKD.² A 2019 study showed that the expression of NOX2 in small-for-gestational-age (SGA) newborn pigs was unchanged compared to the control group, whereas NOX4 expression was significantly elevated. This suggests that early oxidative stress my contribute to the onset of kidney injury in SGA infants.

NOX2 was shown to affect around 500 biochemical pathways. For example, evidence suggests that elevated NOX2 expression may be linked to increased microglial activation, oxidative stress and inflammatory gene expression, contributing to worsened spinal cord injury.³

Similarly, NOX4 was shown to affect around 550 pathways. For instance, silencing of NOX4 has also shown to inhibit NF-κB nuclear translocation, ox-LDL endocytosis and monocytes adhesions in a mouse model.⁴

Using Causaly, a comparative analysis of biochemical pathways between concepts can be performed. From the identified pathways, around 40% were affected by both NOX2 and NOX4, as shown in Figure 2. For example, both NOX2 and NOX4 genes have been associated with cellular senescence, ferroptosis and glycolysis.

Deciphering biochemical pathways influenced by a target offers a better understanding of potential implications arising from modulating these pathways. Human-centric AI can help unravel disease mechanisms and molecular interactions, paving the way for targeted therapies that address the root cause of the disease.