Dunlop, A. | Search bioRxiv

bioRxiv2023

Dietaryindex: A User-Friendly and Versatile R Package for Standardizing Dietary Pattern Analysis in Epidemiological and Clinical Studies

BackgroundFew standardized and open-source tools exist for calculating dietary pattern indexes from dietary intake data in epidemiological and clinical studies. Miscalculations of dietary indexes, with suspected erroneous findings, are occasionally noted in the literature. ObjectiveThe primary aim is to develop and validate dietaryindex, a user-friendly and versatile R package that standardizes the calculation of dietary indexes. MethodsDietaryindex utilizes a two-step process: an initial calculation of serving size for each food and nutrient category, followed by the calculation of individual dietary indexes. It includes generic functions that accept any preprocessed serving sizes of food groups and nutrients, with the standard serving sizes defined according to the methodologies used in well-known prospective cohort studies. For ease of use, dietaryindex also offers one-step functions that directly reference common datasets and tools, including the National Health and Nutrition Examination Survey (NHANES) and Block Food Frequency Questionnaire, eliminating the need for data preprocessing. At least two independent researchers validated the serving size definitions and scoring algorithms of dietaryindex. ResultsDietaryindex can calculate multiple dietary indexes of high interest in research, including Healthy Eating Index (HEI) - 2020, Alternative Healthy Eating Index 2010, Dietary Approaches to Stop Hypertension Index, Alternate Mediterranean Diet Score, Dietary Inflammatory Index, American Cancer Society 2020 dietary index, and Planetary Health Diet Index from the EAT-Lancet Commission. In our validation process, dietaryindex demonstrated full accuracy (100%) in all generic functions with two-decimal rounding precision in comparison to hand-calculated results. Similarly, using NHANES 2017-2018 data and ASA24 and DHQ3 example data, the HEI2015 outputs from dietaryindex aligned (99.95% - 100%) with results using the SAS codes from the National Cancer Institute. ConclusionsDietaryindex is a user-friendly, versatile, and validated informatics tool for standardized dietary index calculations. We have open-sourced all the validation files and codes with detailed tutorials on GitHub (https://github.com/jamesjiadazhan/dietaryindex).

bioinformatics↗

bioRxiv2022

Benchmarking software to predict antibiotic resistance phenotypes in shotgun metagenomes using simulated data

BackgroundThe use of shotgun metagenomics for AMR detection is appealing because data can be generated from clinical samples with minimal processing. Detecting antimicrobial resistance (AMR) in clinical genomic data is an important epidemiological task, yet a complex bioinformatic process. Many software tools exist to detect AMR genes, but they have mostly been tested in their detection of genotypic resistance in individual bacterial strains. Further, these tools use different databases, or even different versions of the same databases. Understanding the comparative performance of these bioinformatics tools for AMR gene detection in shotgun metagenomic data is important because this data type is increasingly used in public health and clinical settings. MethodsWe developed a software pipeline, hAMRoaster (Harmonized AMR Output compAriSon Tool ER; https://github.com/ewissel/hAMRoaster), for assessing accuracy of prediction of antibiotic resistance phenotypes. For evaluation purposes, we simulated a highly resistant mock community and several low resistance metagenomic short read (Illumina) samples based on sequenced strains with known phenotypes. We benchmarked nine open source bioinformatics tools for detecting AMR genes that 1) were conda or Docker installable, 2) had been actively maintained, 3) had an open source license, and 4) took FASTA or FASTQ files as input. hAMRoaster calculated sensitivity, specificity, precision, and accuracy for each tool, comparing detected AMR genes to susceptibility testing. ConclusionOverall, all tools were precise and accurate at all genome coverage levels tested (5x, 50x, 100x sequenced bases / genome length) in the highly resistant mock community with more variability in the low resistance community (1x coverage). This study demonstrated that different bioinformatic tools and pipelines yield differences in AMR gene identification across drug classes, and that these differences become important if researchers are interested in resistance to specific drug classes. SignificanceSoftware selection for metagenomic AMR prediction should be driven by the context of the clinical/research questions and tolerance for true and false negative results. The ability to assess which bioinformatics tool best fits a particular dataset prior to beginning a large-scale project allows for more efficient processing and analysis using optimal tools for a particular research question. As prediction software and databases are in a state of constant refinement, the approach used here--creating synthetic communities containing taxa and phenotypes of interest along with using hAMRoaster to assess performance of candidate software--offers a template to aid researchers in selecting the most appropriate strategy at the time of analysis.

bioinformatics↗