Paper reading (三十三)：Personalized Nutrition by Prediction of Glycemic Responses（Experimental）

本文链接：https://blog.csdn.net/wxw060709/article/details/102537237

正文部分内容摘录：

4. Experimental Procedures

4.1 Human cohorts

Observational Study of Blood Glucose Levels and Gut Microbiota in Healthy Individuals
跟这个human cohorts相关的publications还有一篇：Observational Study of Blood Glucose Levels and Gut Microbiota in Healthy Individuals
Exclusion Criteria:Under 18 years of age；Mental incompetence 所以说没有18岁以下的未成年人了，那这个能代表以色列的整体population吗？还是说影响也甚微？比如说未成年人本身血糖问题也不大。

4.2 Study design

4.3 Standardized meals

Participants were instructed to consume these meals immediately after their night fast, not to modify the meal, and to refrain from eating or performing strenuous physical activity before, and for 2 hr following consumption.

4.4 Stool sample collection

4.5 Genomic DNA extraction and filtering

4.6 Microbial analysis

4.7 Associating PPGRs with risk factors and microbiome profile

We also calculated the mean PPGR of replicates of each standardized meal (if performed) and correlated (Pearson) these values with (a) blood tests; (b) anthropometric measurements; (c) 16S rRNA RA at the species to phylum levels; (d) MetaPhlAn tag-level RA; and (e) RA of KEGG genes.
We capped RA at a minimum of 1e-4 (16S rRNA), 1e-5 (MetaPhlAn) and 2e-7 (KEGG gene). For 16S rRNA analysis we removed taxa present in less than 20% of participants.
Correlations on RAs were performed in logspace.

4.8 FDR correction

4.9 Meal preprocessing

We merged meals logged less than 30 min apart and removed meals logged within 90 min of other meals. We also removed very small (<15 g and <70 Calories) meals and meals with very large (>1 kg) components, meals with incomplete logging and meals consumed at the first and last 12 hr of the connection week.

4.10 PPGR predictor

The depth of the tree at each estimator was not limited, but leaves were restricted to have at least 60 instances (meals).
We used 4000 estimators with a learning rate of 0.002.

4.11 Microbiome changes during dietary intervention

We determined the significantly changing taxa of each participant by a Z test of fold-change in RA between the beginning and end of each intervention week against a null hypothesis of no change and standard deviation calculated from at least 25-fold changes across the first profiling week (no intervention) of corresponding taxa from all participants with similar initial RA.
We checked whether a change was consistent across the cohort for each taxa by performing Mann-Whitney U-test between the Z statistics of the “good” intervention weeks and those of the “bad” intervention weeks across all participants.