Obesity stands as a significant public health concern. Citing research by Lavie and others, there’s been a noticeable uptick in obesity rates across the United States and much of the Western world in recent years, leading to a global epidemic. Obesity aggravates several risk factors for cardiovascular diseases (CVD), consequently elevating the incidence of conditions such as hypertension, coronary artery disease, heart failure, and atrial fibrillation [1][2]. This study seeks to unravel the genetic contributors to obesity by analyzing DNA methylation in saliva samples from Latino preschool children, which could pave the way for precise intervention strategies including specialized medications and metabolic adjustments.
The DNA methylation profiles in saliva from 96 mother-child pairs were obtained from the NCBI Gene Expression Omnibus (GEO) database (Click here), with the accession code GSE72556. This methylation data was acquired using the Illumina HumanMethylation450 BeadChip platform.
The dataset comprises two text files: a dictionary linking CpG sites to genes, and the main data file with 96 samples and over 480,000 CpG site columns. We utilized the data.table package for efficient data handling. Three samples were excluded due to participant withdrawal. Cluster analysis identified and removed an outlier. We then discarded columns with over 20% missing values to enhance model accuracy and imputed remaining missing data using the KNN algorithm.
KNN imputation with K = 10 was optimal, as shown in figure 1, resulting in a final dataset of 92 observations and 364,230 CpG sites. Children were categorized into obese and normal groups based on the International Obesity Task Force (IOTF) BMI cutoffs (Details here), where a 3 years old boy with BMI less than 17.69 is classified as normal, etc.. The division according to IOTF indexes resulted in 25 obese and 67 normal-weight children.
K value vs. Number of Genes selected by Elastic Net
We use BH (Benjamini-Hochberg) method and cutoff equals 0.1 to find the best alpha value. Figure 2 indicates that an α value of 0.5 in Elastic Net maximizes pathway identification in KEGG analysis and the effect of different alpha values on GO enrichment results. In order to get results from both enrichment analysis, we select α = 0.5 for future analysis.
Alpha value vs. Number of enriched terms. KEGG on the left, GO on the right
Then, we use 10-fold corss validation to find the best lambda. Figure 3 aids in selecting the most appropriate lambda value by using 10-fold cross validation, correlating it with residuals in the Elastic Net model (best λ = 0.167).
ten fold cross validation to select best lambda value
Elastic Net Regularization was applied for predictor selection, followed by KEGG and GO enrichment analysis to identify gene patterns. The table of Elastic net regression is shown below. The Gene column are provided by the dictionary provided along with the dataset.
Enrichment analyses yielded various pathways: KEGG analysis pointed to protein digestion and absorption (p.adjust = 0.0897), while GO analysis identified actin filament binding, transmembrane transporter binding, actin binding, and voltage-gated potassium channel activity (p.adjust between 0.1 and 0.05).
results of Enrichment analysis. KEGG on the left, GO on the right
The methylation status of CpG islands might affect the expression of enzymes related to protein digestion, such as pancreatic proteases or gastric proteases. In children, changes in the activity of these enzymes might impact the effective breakdown of proteins and absorption of amino acids, thereby affecting growth and development. Altering the methylation status of specific genes might change the pathways of protein metabolism and its long-term impact on child health.
Actin filaments play a key role in cell morphology and movement, functions that are also important in intracellular transport and cell signaling. Changes in actin-binding proteins may affect the internal transport of substances in cells, including the transport of nutrients, potentially impacting cellular metabolic activities and energy balance.
Transmembrane transporters are responsible for the transport of substances inside and outside the cell, including nutrients, metabolic wastes, and signaling molecules. This suggests that methylation changes might affect cellular uptake and metabolism of nutrients, especially in energy balance and the development of obesity.
Voltage-gated potassium channels play a key role in maintaining cell potential and signal transduction. Changes in the activity of these channels might affect the functioning of the nervous system, including the regulation of appetite and energy expenditure, an important aspect in obesity research.
The Elastic Net analysis of saliva samples in our study has notably aligned with existing literature on PubMed, reaffirming the reliability of this method in identifying CpG islands associated with obesity and protein metabolism. This consistency underscores the effectiveness of Elastic Net in pinpointing potential epigenetic markers in children. Such markers play a crucial role in understanding how genetic predispositions to obesity and metabolic disorders manifest in early childhood.
While Elastic Net provides a robust framework for our analysis, it’s imperative to acknowledge certain limitations. The size of our sample, although adequate for preliminary findings, may limit the generalizability of our results. Additionally, the statistical power of our study could be further enhanced with a larger cohort, offering more definitive conclusions. There’s also the potential for biases inherent in any observational study, which should be considered when interpreting our findings.
Our research paves the way for future studies to delve deeper into the relationship between specific methylation patterns of CpG islands and various nutritional statuses. By extending this research to include children from diverse age groups and backgrounds, we can gain a more comprehensive understanding of how these epigenetic patterns vary or remain consistent across different demographics. Such studies could profoundly impact our understanding of childhood nutrition and health risks, potentially influencing public health policies by providing evidence-based guidelines for dietary and lifestyle modifications.
One of the most promising aspects of our study is the use of saliva as a non-invasive medium for identifying potential biomarkers of obesity and metabolic dysfunction [3]. The ease of collecting saliva makes it a highly accessible tissue for large-scale screenings and longitudinal studies. This could be instrumental in the early detection of children at risk, offering a window for timely intervention and possibly altering the trajectory towards obesity and related metabolic disorders.
In conclusion, our study not only reaffirms the utility of Elastic Net in epigenetic research but also opens new avenues for understanding and combating childhood obesity. By leveraging saliva samples for non-invasive testing and focusing on early-life epigenetic changes, we stand at the cusp of a transformative approach in addressing one of the most pressing public health challenges of our time. The potential for these findings to inform public health policies and individualized interventions could have lasting impacts on the health trajectories of children globally.
Liancheng Lu: was responsible for the finding of dataset, part of the project codes. Haoyi Zheng: was resonsible for the plotting of the project codes. Yue Yu: was responsible for the presentation and the final report.
The work directory of this entire project has been published on a Github repository, which can be accessed here: Click here.
-
Carl J. Lavie, Alban De Schutter, Parham Parto, Eiman Jahangir, Peter Kokkinos, Francisco B. Ortega, Ross Arena, Richard V. Milani, Obesity and Prevalence of Cardiovascular Diseases and Prognosis—The Obesity Paradox Updated, Progress in Cardiovascular Diseases, Volume 58, Issue 5, 2016, Pages 537-547, ISSN 0033-0620, https://doi.org/10.1016/j.pcad.2016.01.008. (https://www.sciencedirect.com/science/article/pii/S0033062016300081)
-
Ortega FB, Lavie CJ, Blair SN. Obesity and cardiovascular disease. Circ Res. 2016. doi: 10.1161/CIRCRESAHA.115.306883. (https://doi.org/10.1161/CIRCRESAHA.115.306883)
-
Oelsner, Kathryn Tully; Guo, Yan, To; Sophie Bao-Chieu; Non, Amy L.; Barkin, Shari L.. 2017, 2017/01/09. Maternal BMI as a predictor of methylation of obesity-related genes in saliva samples from preschool-age Hispanic children at-risk for obesity. BMC Genomics. https://doi.org/10.1186/s12864-016-3473-9. Doi: 10.1186/s12864-016-3473-9.
