Navigating the bridge between wet and dry lab toxicology research to address current challenges with high-dimensional data

Abstract

Toxicology research has rapidly evolved, leveraging increasingly advanced technologies in high-throughput approaches to yield important information on toxicological mechanisms and health outcomes. Data produced through toxicology studies are consequently becoming larger, often producing high-dimensional data. These types of data hold promise for imparting new knowledge, yet inherently have complexities causing them to be a rate-limiting element for researchers, particularly those that are housed in “wet lab” settings (i.e., researchers that use liquids to analyze various chemicals and biomarkers as opposed to more computationally focused, “dry lab” researchers). These types of challenges represent topics of ongoing conversation amongst our team and researchers in the field. The aim of this perspective is to i) summarize hurdles in analyzing high-dimensional data in toxicology that require improved training and translation for wet lab researchers, ii) highlight example methods that have aided in translating data analysis techniques to wet lab researchers; and iii) describe challenges that remain to be effectively addressed, to date, in toxicology research. Specific aspects include methodologies that could be introduced to wet lab researchers, including data pre-processing, machine learning, and data reduction. Current challenges discussed include model interpretability, study biases, and data analysis training. Example efforts implemented to translate these data analysis techniques are also mentioned, including online data analysis resources and hands-on workshops. Questions are also posed to continue conversation in the toxicology community. Contents of this perspective represent timely issues broadly occurring in the fields of bioinformatics and toxicology that require ongoing dialogue between wet and dry lab researchers.

Wildfires and climate justice: future wildfire events predicted to disproportionally impact socioeconomically vulnerable communities in North Carolina

“…It is notable that the process of data selection itself is also a source of potential bias, and authors aim to update findings upon new data releases that more holistically capture socioeconomic vulnerability as the field continues to advance. Additionally, the usage of clustering, which is an unsupervised machine learning approach ( 40 ), was leveraged to determine co-occurrence patterns between SES variables and wildfire risk. It is notable that these methods do not capture causal relationships; rather, the selected methods represent unbiased pattern recognition approaches that are inherently exploratory, setting the stage for future research to further quantify likelihoods and future impacts of wildfire events alongside factors of climate change.…”

“…To identify distinct regions throughout NC based upon SES profiles, k -means clustering was employed. Prior to analysis, data were standardized using the scale function in base R. K -means, an unsupervised machine learning technique ( 39 , 40 ), was used to find patterns between census tracts based on SES data distributions. Cluster assignments were derived by minimizing within cluster differences and maximizing between cluster differences using the factoextra package in R ( 41 ).…”

Burn Pit Smoke Condensate-Mediated Toxicity in Human Airway Epithelial Cells

“…This subset of sample conditions was prioritized to enhance the resulting visualizations while focusing on conditions of high importance. PCA is an unsupervised machine learning technique that seeks to compress the variance from an original data set into the fewest number of principal components. , PCA findings can be visualized via plots, wherein each axis represents a component that captures large amounts of variance in the data; in this case, the first two principal components were plotted within a two-dimensional (2D) plot to visualize potential patterns among sample conditions. PCA was performed using the factoextra package.…”

“…To do so, these expression profiles were visualized using a heat map showing average concentrations for each biomarker after exposure to each sample condition using the pheatmap package. The biomarkers in the heat map were grouped together using hierarchical clustering to group together cytokines or genes based on the average distance between them . This clustering approach was implemented to identify groups of cytokines with concerted response profiles across exposure conditions, as previously described. , …”

“…The biomarkers in the heat map were grouped together using hierarchical clustering to group together cytokines or genes based on the average distance between them. 37 This clustering approach was implemented to identify groups of cytokines with concerted response profiles across exposure conditions, as previously described. 41,42 Correlation Analysis between Polyaromatic Hydrocarbons (PAHs) and Cytokine Levels.…”