Molecular signatures of food intake in human stool


Dietary intake is one of the most complex and sustained environmental exposures of the human lifespan. At present, field standard methods for dietary assessment all depend on asking people to self-report what they eat. Self-reported dietary data have enabled the study of diet-health relationships and informed decades of nutritional policy, but they are also limited in their ability to collect data from individuals across the spectrum of age, cognitive ability, and culture. To address the need for alternative measures of diet, we leverage DNA sequencing technology to measure diet from degraded food DNA in human stool samples. This strategy, termed “DNA metabarcoding,” works by amplifying and sequencing regions of the genome to identify foods by their DNA sequence. Although metabarcoding strategies have been applied in other diet-related fields, they have never been used in nutritional epidemiology.

In Chapter 1, I describe the potential of DNA metabarcoding for human dietary assessment. In Chapter 2, we apply DNA metabarcoding with the chloroplast trnL-P6 marker region to >1,000 stool samples from >300 unique participants. Across multiple cohorts, we find that the number of plant taxa detected per sample by trnL metabarcoding correlates with dietary diversity and quality estimates from established self-report assessments. In a cohort where the administration of self-reports failed, we use trnL metabarcoding to retrospectively derive diet data, and detect relationships between trnL dietary diversity and age, socioeconomic indicators, and health that replicate prior epidemiologic findings.

In Chapter 3, we perform a pilot study of a second molecular technique— metaproteomics— to determine if it can address a key limitation of metabarcoding: the inability of DNA to distinguish food type for foods like chicken and egg, which are derived from the same source species. We identify tissue-specific protein signatures of food intake and confirm an overall correspondence between DNA- and protein- based dietary assessment in comparison to written records. We also expand our metabarcoding protocols to include the mitochondrial 12SV5 marker for combined detection of plant and animal intake. Taken together, these findings establish the value of molecular measurements from stool as a strategy to monitor and improve nutrition.





Petrone, Brianna Leigh (2024). Molecular signatures of food intake in human stool. Dissertation, Duke University. Retrieved from


Dukes student scholarship is made available to the public using a Creative Commons Attribution / Non-commercial / No derivative (CC-BY-NC-ND) license.