Back to Journal

SM Journal of Environmental Toxicology

Elja: An R Package to Perform Environment-Wide Association Studies (Ewas / Envwas) Analysis

Abstract Citation Introduction Methods Aim of the Package Implementation and Design Application Example Conclusion Funding References
Details

Received: 17-Mar-2024

Accepted: 23-Mar-2024

Published: 25-Mar-2024

Marwan El Homsi¹* and Isabella Annesi-Maesano¹,²

¹Desbrest Institute of Epidemiology and Public Health, Univ Montpellier, INSERM, Montpellier, France
²Division of Respiratory Medicine, Allergology, and of Thoracic Oncology, University Hospital of Montpellier, Montpellier

Corresponding Author:

Marwan El Homsi, Desbrest Institute of Epidemiology and Public Health, Univ Montpellier, INSERM, Montpellier, France

Keywords

Epidemiology; Exposome; R; EnvWAS; EWAS; Package

Abstract

Environment-Wide Associations Studies (EWAS / EnvWAS) are an approach involving multiple analyses of the relationship between several exposures to a health event. This approach allows us to study the impact of an exposome on a health event. The Elja package was developed using R to perform the EnvWAS analytical approach. This package performs repeated analyses in the case of linear regression models, logistic regression models, and, more generally, generalized linear models.

The results are presented in two complementary ways. In a detailed way, with a table automatically generated at the end of the analysis allows all results to be consulted and then extracted. In a more visual way, with a Manhattan plot displaying the results in their entirety, with a significance threshold of 0.05 and the corrected threshold.

This package is compatible with Windows, Ubuntu and MacOS. It is available free of charge on GitHub (https://github.com/EHMarwan/ Elja) and the CRAN directory.

Citation

El Homsi, Annesi-Maesano I.(2024)Elja: An R Package to Per form Environment-Wide Association Studies (Ewas / Envwas) Analysis. SM J Environ Toxicol 7: 4.

Introduction

The impact of environmental factors on human health is a major public health concern. The most common studies have attempted to estimate the degree of association between an isolated environmental factor and a health event. However, this approach is insufficient for explaining multifactorial diseases. The need to consider the exposome–that is, all the exposures to which an individual is subjected over the course of his or her life–has emerged [1,2]. To achieve this, a new approach has been developed that aims to consider several environmental factors at once in relation to a health event in order to understand their overall impact on this event. The exposome includes all external factors, including chemical, physical, biological, and social elements, which can potentially influence an individual’s health and well-being. In the world of exposome studies, one methodology is increasingly in use: Environmental-Wide Association Studies (EWAS, later called EnvWAS).

The first studies using the EnvWAS approach were conducted in 2010, with two articles introducing this new approach to study the association between the exposome and a health event, namely diabetes [3,4]. This approach aims to study the association between a complex health issue and several environmental exposure factors. In this approach, multiple environmental exposures are analyzed in relation to the same health event, incorporating methods for controlling alpha risk inflation, which are common when multiple tests are performed. The advantage of this approach is that it enables us to study the association of several environmental factors simultaneously in relation to a health event, and thus to identify new statistically significant protective or risk environmental factors for this event. The use of the EnvWAS could lead to a better understanding of the effects of environments with a wider angle of vision (urban, work, etc.) on people’s health in relation to certain environmentally sensitive conditions. Moreover, unlike traditional epidemiological studies that focus on one or a few exposures at a time, the EnvWAS adopts a more comprehensive approach by simultaneously considering multiple environmental factors. This allows researchers to explore the cumulative effects and interactions between different exposures, thus providing a more holistic view of the exposome.

The EnvWAS approach and the creation of codes for the analysis and presentation of results can be complex; therefore, a package is useful to simplify the entire process and make it more accessible. The Elja package simplifies the EnvWAS process and yields tabular and graphical results. Recently developed packages on R, not available on CRAN, have also been developed for this purpose (25,26).

Methods

In the Elja package, EnvWAS employs advanced statistical methods, such as linear regression models, logistic regression models, and generalized linear models, to analyze the complex relationships between exposures (independent variables) and health outcomes (dependent variable). The Akaike Information Criterion (AIC) was used to select the most appropriate model by considering both goodness of fit and model complexity [5], the Bonferroni-corrected alpha threshold [6], and the Benjamini-Hochberg False Discovery Rate (FDR) [7] were used to adjust the significance level in hypothesis testing when multiple comparisons are made simultaneously.

Aim of the Package

Simplification of the EnvWAS approach

The first objective of this package is to simplify the implementation of EnvWAS. The analytical approach consists of creating loops that, through an adapted statistical model, analyze the effect of an exposure (independent variable) on a health event (dependent variable). Each exposure is considered for each model using a loop. This package allows the complete EnvWAS process to be carried out by defining at least two elements: the targeted health event in a table (dataframe) containing, among other variables, the exposures whose association is to be tested.

The visual impact of this package

The Elja package offers two types of outputs. A data frame is provided for each variable tested (as well as all the modalities of these variables), including the value of the estimator (odd ratio or coefficients), its 95% confidence interval, the associated p-value, the number of values considered in the model, and the Akaike Information Criterion (AIC) of the model. In addition, two Manhattan plots can be displayed, both with a visual indicator of the alpha threshold of 0.05, the Bonferroni-corrected alpha threshold, and the Benjamini-Hochberg False Discovery Rate (FDR). The first one, representing all the variables in the EnvWAS analysis. The second one, with only significant values.

Implementation and Design

Installation

The Elja package can be installed from GitHub using the devtools [8] package as follows:

install.packages(“devtools”)

library(devtools)

install_github(“Displayr/flipPlots”)

It is also available in the CRAN directory and can be installed as follows.

install.packages(“Elja”)

This package is publicly available and compatible with R for Windows, MacOS, and Ubuntu. It also requires the stats [9], base [9], MASS [10] et dplyr [11] packages and for the graphics aspect, these were created using the ggplot2 package [12].

Main features

Using a data frame containing data on the health event and exposure factors to be studied, it is possible to perform an EnvWAS-type analysis by defining the health event of interest that will be used for the analysis within the function specific to the type of model required. The result is represented in the form of a table with the possibility of also having an automatically generated Manhattan plot. Several functions will be added in response to community needs.

The functions included in the package to perform EnvWAS analysis and present the results follow the same approach. The first step is to construct the equations for each model from the dataset and the instructions provided. The second step is to run each model in a loop and store the results. Finally, the results are displayed in a table and, if required, in a graphical form. This package is designed for linear regression, logistic regression, and generalized linear models called ELJAlinear, ELJAlogistic, and ELJAglm, respectively. The functional approach of the function contained in this package is shown in figure 1.

Figure 1: Graphical representation of the functional flow from input to output of the functions in the Elja package.

Application Example

Presentation of the data set

This example used the PIMA dataset [13] extracted from the mlbench package [14]. The PIMA dataset was obtained from the National Institute of Diabetes and Digestive and Kidney Diseases. The aim of the dataset is to predict whether a patient has diabetes based on certain individual characteristics included in the dataset. This dataset contains only women aged at least 21 years and of Pima Indian origin. This dataset contained the following variables:

• pregnant: Number of times pregnant

• glucose: Plasma glucose concentration a 2 hour in an oral glucose tolerance test

• pressure: Diastolic blood pressure (mm Hg)

• triceps: Triceps skin fold thickness (mm)

• insulin: 2-Hour serum insulin (μU/ml)

• mass: Body mass index (weight in kg/(height in m)2)

• pedigree: Diabetes pedigree function

• age: Age (years)

• diabetes: test for diabetes (pos/neg)

Use of the Elja Package with PIMA Data

In this example, we will use the “diabetes” variable as a health event and all other variables as exposures. Since the outcome is binary categorical, we use a logistic regression model with the ELJA logistic function. The results of the analysis are presented in table form (table 1) and Manhattan plot (figure 2). A Manhattan plot with significant results is not displayed here.

Figure 2: Manhattan plot showing the results of EnvWAS analysis with a logistic regression model on PIMA data, with diabetes as the outcome variable. The p-values are shown on the abscissa. The horizontal bars refer to the different estimated alpha thresholds: in yellow the 5% alpha threshold, in purple the alpha threshold corrected according to Benjamini-Hochberg’s False Discovery Rate method, and in black the alpha threshold corrected according to Bonferroni’s method.

Conclusion

The Elja package represents a new tool designed to simplify the coding of EnvWAS, and the representation of the results obtained. Its use can enable the analysis of large datasets to determine the impact of environmental factors on health events. We believe that this package can simplify the exploitation of datasets involving several factors related to the same health event. The Elja package is publicly available on CRAN and GitHub.

Funding

This study was funded by the EU-funded URBAN Observatory for Multi-participatory Enhancement of Health and Wellbeing (URBANOME) Project (https://www.urbanome.eu/), EU HORIZON 2020 Grant #945391. Marwan EL HOMSI is a recipient of an EU PhD fellowship within the URBANOME project.

Availability and Requirements

• Project name: Elja

• Project home page: https://github.com/EHMarwan/Elja

• Operating system(s): Windows, MacOS and Ubuntu

• Programming language: R

• Other requirements: stats, devtools, dplyr, ggplot2, MASS are required.

• License: GPL-3

References

1. Miller GW, Jones DP. The nature of nurture: refining the definition of the exposome. Toxicological Sciences. 2014; 137: 1-2.

2. Niedzwiecki MM, Walker DI, Vermeulen R, Chadeau-Hyam M, Jones DP, Miller GW. The exposome: molecules to populations. Annu Rev Pharmacol Toxicol. 2019; 59: 107-127.

3. Thomas D. Gene–environment-wide association studies: Emerging approaches. Nat Rev Genet. 2010; 11: 259-272.

4. Patel CJ, Bhattacharya J, Butte AJ. An Environment-Wide Association Study (EWAS) on type 2 diabetes mellitus. PLOS ONE. 2010; 5: e10746.

5. Akaike H. A new look at the statistical model identification. IEEE Transactions on Automatic Control. 1974; 19: 716-723.

6. Dunn OJ. Multiple comparisons among means. Journal of the American Statistical Association. 1961; 56: 52-64.

7. Benjamini Y, Hochberg Y. Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B (Methodological). 1995; 57: 289-300.

8. Wickham H, Hester J, Chang W, Bryan J. devtools: Tools to Make Developing R Packages Easier. 2023.

9. R Core Team. R: A Language and Environment for Statistical Computing. 2023.

10. Venables WN, Ripley BD. Modern Applied Statistics with S. Springer. 2002.

11. Wickham H, François R, Henry L, Müller K, Vaughan D. dplyr: A Grammar of Data Manipulation. 2023.

12. Wickham H. Ggplot2. Springer International Publishing. 2016.

13. Smith JW, Everhart JE, Dickson WC, Knowler WC, Johannes RS. Using the ADAP learning algorithm to forecast the onset of diabetes mellitus. Proc Annu Symp Comput Appl Med Care. 1988: 261-265.

14. Newman D, Hettich S, Blake C, Merz C. UCI Repository of machine learning databases. 1998.

Other Articles

Article Image 1

Ecological Models for Predicting Contaminant Effects

Anthropogenic and natural environmental contaminants are a common problem and a source of concern to ecosystem health. Industrial toxins are one of the leading causes of pollution worldwide. Industrial toxins may arise as a result of air emissions, water releases, water seepage, air deposition or disposal and leaching of solid waste. The combination of natural and anthropogenic sources of toxins present challenges with respect to the protection of local ecological environments. Predicting the impacts of environmental contaminants on ecosystems become an important part of the decision-making process for managing environment problems. To protect ecological environments and species, it is necessary to assess the risk to organisms exposed to toxins, and find relevant factors that determine the persistence and extirpation of populations. Over the past several decades, ecotoxicological models have been widely applied to predict contaminant effects.

Qihua Huang*


Article Image 1

Solubilization of Metal Particles and Lung Toxicity

Inhaled particles are readily phagocytosed by alveolar macrophages (AMs) present in the lung. Other routes of particle exposure targeting the lung and frequently used with animal models include intratracheal instillation and oropharyngeal aspiration. The mechanism (s) by which AMs drive pulmonary toxicity downstream of particle uptake is not fully understood, as well as the contributing role of other phagocytic cell types present in the airways and lung including epithelial cells.

William M Gwinn*


Article Image 1

Changes in the Pituitary and hypothalamus Monoaminergic Neurotransmitters after Acute and Prolonged Stress Exposure to Benzo (α) Pyrene in Acanthopagrus latus

In this study, the effects of the Polycyclic Aromatic Hydrocarbon (PAH) Benzo (α) pyrene (BαP) exposure on the levels of serotonin (5-HT) and dopamine (DA) in the pituitary and hypothalamus of Yellowfin seabream, (Acanthopagrus latus) were examined. To assess the acute stress responses, vegetable oil (2 µl g-1) containing BαP (50 mg kg-1) was injected into the treatment group of fish (the vegetable oil alone was the control), and brain samples from different groups were collected 3 hr after injection. Base line group was not injected. To study the long-term stress, brains were collected from both injected groups after 72 hr. The amounts of serotonin, dopamine, and amine metabolites in the hypothalamus and pituitary were measured. Results showed that BαP influenced the serotonergic system more than the dopaminergic system for both acute and prolonged stress in both the hypothalamus and pituitary. Acute exposure to BαP induced significant decreases in DA and increases in DOPAC (3, 4-dihydroxyphenylacetic acid) in the pituitary (P<0.05). Major changes induced by both acute and prolonged exposure to BαP included significant decreases in 5-HT, increases in 5-HIAA (5-hydroxy-3- indoleacetic acid) and increases in the 5-HIAA/5-HT ratio (P<0.05). These exposures might affect the synthesis, storage, uptake/release, and degradation of the neurotransmitters in the hypothalamus and pituitary of sea bream, especially the pituitary gland.

Sara Rastgar, Abdol-Ali Movahedinia, Ahmad Savari, Hosein Pasha Zanosi, Morteza Behnam Rasoli, and Rashid Alijani Ardeshir*


Article Image 1

Policy-Practice Nexus: Pesticide Registration, Distribution and Use in Ethiopia

To promote environmental governance of pesticides, Ethiopia has developed a legal framework on pesticide registration, distribution and use. However, there is no clear answer to the question whether the policies on pesticide registration, distribution and use were implemented in an effective and sustainable way at the national and local levels. Therefore, this study is designed to assess the gap between the state pesticide policy and its implementation with respect to pesticides registration, distribution and use in Ethiopia. The data were collected from state’s pesticide experts (regulators), distributors (importers and retailers) and end users (vegetable framers) through structured and semi structured interview, observations and existed documents. The data were analyzed in combination of qualitative and quantitative methods. The empirical results indicated that existing law do not function in an adequate way due to inefficient implementation and missing legal instruments. Thus, present shortcomings and future endeavors may need to find ways to envisage better implementation of the law that is designed to govern pesticide use by farmers from registration to distribution and use and monitoring, including quality control. Last but not least, there is a need to increase capacity of state actors (human, financial and material) both in the national and local level and participation of private actors should be considered for sustainable pesticide governance that will benefit Ethiopia.

Belay Tizazu Mengistie*


Article Image 1

Biodegradation of Toxic Compounds by Aerobic Granulation Technology

Aerobic granulation technology is a strong contender to replace conventional wastewater treatment processes. Aerobic granules are microspheres of self-immobilized microorganisms and they can be easily separated from the liquid phase as they settle many times faster than sludge flocs. Biodegradation of toxic compounds is a crucial area where traditional wastewater treatment processes are still struggling. It is a relatively new research area for aerobic granulation technology, but it holds the key for this technology to establish itself as an effective alternative of present day wastewater treatment processes. This review covers the mostly explored toxic compounds which have been successfully treated by aerobic granulation technology. Present research trends of this technology in biodegradation of toxic compound have been critically analyzed to identify the future research directions. In addition to a summary of different reports, authors’ own opinions on this subject have been presented

Saurabh Jyoti Sarma¹ and Joo-Hwa Tay¹*


Article Image 1

Participatory Demonstration of Gully Treatment Method for Gully Rehabilitation in Dollo Schem, Kamba District, South Ethiopia

Gully erosion is the major environmental problem threating huge area of agricultural lands in south Ethiopia, particularly. The present study aimed at evaluating demonstrates gully treatment methods for gully rehabilitation in Kamba, South Ethiopia. On farm filed experiments, key informant interview and focus group discussion were used to collect and analyze experimental data. The result of this study indicates that integration of check dame with Elephant grass have significant effect on gully rehabilitation. This integration was reduced gully depth, slope gradient, gully volume and soil loss. Slope of gully bed was reduced by 60%, 46% 37% respective years. Similarly gully depth reduced by 2.54m, 1.72m and 1.456m while gully volume was reduced by 37.44m3, 8.81m3 and 4.32m3 due to check dam as the same order of slope changes. Data such as sediment deposition and biomass production were collected, to investigate their effectiveness in reducing soil erosion and biomass production. Over a period of two years practices caused measurable gully morphology resulted soil calculated was 248.41kg of soil were deposited in 12.76m2 area. According to Focus group discussion farmers viewed these gully rehabilitation measures was positively, apart from the high labour, input materials and technical requirements of stone check dam. Based on the results the study concluded that check dam integrated with Elephant grass in the gully bed have positive effective rehabilitate gully erosion in the study area. Therefore, farmers should use gully rehabilitation techniques to preserve their land.

Amare Gojjam¹*, Yenealem Gemi¹, Tademe Minase¹, Dagnaw Ademe¹ and Birhanu Wolde¹


Article Image 1

Growing of Ipomoea aquatica in Seaweed Supplemented Hydroponic System and Identifying Possible Genera of Microbes in the System

Hydroponic system is method of growing plant in nutrient rich solution. The seaweed biomass that is usually disposed as waste once the carrageenan component is extracted from it can be used as a fertilizer. Since nutrients in hydroponic system provide an ideal environment for microorganisms to grow, water-based microorganisms can be easily introduced into the system. This study compared the growth of water spinach (Ipomoea aquatica) grown in hydroponic system. The effects on the growth of plants in the hydroponic system when seaweed waste is used as the fertilizer were also studied. Furthermore, the possible genera of the microbes present in the water samples obtained from the hydroponic system were identified. The plants were grown in treated hydroponic tank (with seaweed waste fertilizer) and untreated hydroponic tank (without seaweed waste fertilizer). The mean heights of the plants in each condition were obtained and compared. Water samples were collected from the treated and untreated and biochemical tests were also carried out using the water samples collected to identify the possible genera of the microbes present in the hydroponic system. The plants height of treated tank is significantly taller than untreated tank. The microbe present in the treated tank was either Chromobacterium spp./ Vibrio spp./ Plesiomonas spp./ Aeromonas spp., Acinetobacter spp., Bordetella bronchiseptica/ Alcaligenes spp./ Shewanella spp./ Pseudomonas spp.(alkali producers), Kurthia spp. and Cardiobacterium spp., and in the untreated tank were Acinetobacter, Bordetella bronchiseptica / Alcaligenes spp./ Shewanella spp./ Pseudomonas spp. (alkali producers) and Chromobacterium spp./ Vibrio spp./ Plesiomonas spp./ Aeromonas spp. Further research need to be carried out to identify the exact bacterial strain of each microbe. By identifying the microbial population of the hydroponic system, further studies can be carried out to identify the beneficial and harmful bacteria. The harmful bacteria can then be eliminated from the hydroponic system to promote healthy growth of plants through hydroponic cultivation.

Vi Sion Chang and Swee Sen Teo*


Article Image 1

Neurolathyrism - A Case Report and Current Views

Neurolathyrism is caused by excessive consumption of Lathyrus sativus (grass pea) and seen in India and parts of sub-Saharan Africa. The disease manifests as irreversible spastic paraparesis and quadriparesis leading to permanent disability. We report two unique cases of Lathyrism secondary to consumption of Lathyrus sativus and review the literature available. Two brothers from a remote village in India presented with slowly progressive spastic paraparesis of unknown aetiology. On neurological examination, they were found to have motor neuron pattern of weakness. Routine laboratory work-up, CSF studies and imaging of the brain and spine were all found to be unremarkable. Upon further review, it was found that the patients had a prolonged history of consumption of Lathyrus sativus legume. There is no known cure for neurolathyrism. As a result, the legume was banned from staple diet. However, the legume continues to be consumed despite the ban and public education.

Khosa Shaweta¹, Khosa Gurveer S², and Mishra Shri K³


Article Image 1

Biodiversity, Application of Bioenergy for Energy Systems or Materials and Environmental Sustainability: Sustainable Energies Future Perspective through Energy Efficiency Development

The demand for energy continued to outstrip supply and necessitated the development of biomass option. Residues were the most popular forms of renewable energy and currently biofuel production became much promising. Agricultural wastes contained high moisture content and could be decomposed easily by microbes. Agricultural wastes were abundantly available globally and could be converted to energy and useful chemicals by a number of microorganisms. Compost or bio-fertilizers could be produced with the inoculation of appropriated thermophilic microbes which increased the decomposition rate, shortened the maturity period and improved the compost (or bio-fertilizer) quality. The objective of the present research was to promote the biomass technology and involved adaptive research, demonstration and dissemination of results. With a view to fulfill the objective, a massive field survey was conducted to assess the availability of raw materials as well as the present situation of biomass technologies. In the present communication, an attempt had also been made to present an overview of present and future use of biomass as an industrial feedstock for production of fuels, chemicals and other materials. We may conclude from the review paper that biomass technology must be encouraged, promoted, invested, implemented, and demonstrated, not only in urban areas but also in remote rural areas.

Abdeen Mustafa Omer*


Article Image 1

Utilization of two Bacterial Strains (Ochrobactrum Intermedium BC1 and Cupriavidus Taiwanensis LA) to Biodegrade Anthracene, Fluorene, and Naphthalene

Polycyclic aromatic hydrocarbons, known as PAHs, typically persist in the environment, exposing humans to a? considerable health hazard because of the toxins they contain and their capability of triggering cancer. Anthropogenic activities have introduced high levels of PAHs into Arabian Gulf countries’? soil and coastal waters. Numerous studies have indicated that diverse bacterial? strains can successfully break down PAHs. The deduction made is that biodegradation stands as the top choice in terms of? safety, effectiveness, and affordability when it comes to handling PAH-contaminated sites and locations. The efficiency of degrading three PAHs was analyzed in this study with the use? of two novel bacterial strains, considering the optimal temperature and pH requirements. Coastal sediments from the Eastern Province of Saudi Arabia yielded pure cultures of Ochrobactrum intermedium BC1 and Cupriavidus? taiwanensis LA, which were subjected to spiking with 100ppm concentrations each for anthracene, fluorene, and naphthalene. They were then incubated at 25°C in? a shake incubator for 18 days. A solid-phase micro-extraction (SPME) device was used. The extraction of residual PAHs? was the main objective of using the SPME device. Gas Chromatography/Mass Spectrometry (GC-MS) was utilized to quantify and? analyze the residues at predefined time intervals. By the conclusion of the 18-day timeframe, Ochrobactrum intermedium BC1 degraded naphthalene completely. Additionally, anthracene? experienced a reduction of approximately 87%, while fluorene underwent a decrease of about 67. Cupriavidus taiwanensis LA degraded anthracene, fluorene, and? naphthalene by 88%, 53%, and 91% respectively. The degrading efficiency of these novel strains of? bacteria is evidenced by these results. In closing, these strains can be considered potential members of a consortium of microbes? capable of degrading PAHs that can be employed effectively in various cleanup endeavors.

Mordecai J¹, Al-Thukair A²*, Jameel Al-Thagfi³, Basheer Chanbasha³, and Alexis Nzila²