Comments from curator
Phenotyping data comes from
Supplementary Table 3 of the paper (file 41598_2023_41027_MOESM1_ESM.xlsx, sheet "(G) Raw Data", in the
Supplementary Information; also on
PMC). As the Data availability statement says, this sheet holds all the data used in the analyses: one value per DGRP line for 3 metabolites and 9 microbial taxa, for 118 lines. Sheets A-E (GWAS variants with P < 10
-5) and F (gene annotation) are results and are not uploaded.
Design. The DGRP lines (118 in the table; the Methods say 120) were reared on agar-corn-sugar-yeast medium at 25 °C, 65% humidity, 12:12 h light:dark, at similar egg and larval densities.
-
Gut microbiota: one pool of 5 guts dissected from surface-sterilised 5-7-day-old males per line. The loads of total bacteria (16S rRNA), total fungi (ITS), Ascomycota, Basidiomycota,
Acetobacter persici,
A. pomorum,
Enterococcus faecium,
Lactobacillus plantarum and
L. brevis were quantified by SYBR Green qPCR (duplicate reactions, primers in Supplemental Table 2). The bacterial species are expressed as 2
-ΔCt relative to total bacteria and the fungal phyla as 2
-ΔCt relative to total fungi; the paper does not say how the total bacteria and total fungi values were computed.
-
Metabolic pools: one homogenate of 2 flies per line. Glycogen and trehalose (glucose kit after amyloglucosidase or trehalase digestion) and triglyceride were measured in duplicate assays (Cayman kits), averaged, and expressed in mg per mg protein (Bradford). The Methods give no sex or age for these flies; the Discussion says that inbred males of the same age were used, so all values are uploaded with sex M. Glucose was also measured (mean 0.19 ± 0.12 mg/mg protein), but its per-line values are not published.
The values are the authors' normalised values, not mg/mg or relative abundances. As the distributions deviated from normality, the authors normalised each variable with the R package bestNormalize before all their analyses, and only the normalised values are published. All phenotypes are therefore unitless (higher values = more of the metabolite or taxon):
- the 9 microbial loads and trehalose are ordered quantile (rank-based inverse normal) scores: every value equals qnorm((rank - 0.5)/n) over the lines with a value, within 5 × 10
-10 (mean 0, range about ±2.6);
- triglyceride is a standardised monotone transform (mean 0, SD 1) that is not rank-based (it fits a Box-Cox-type power transform of the mg/mg values);
- glycogen is not centred (mean -1.38, range -3.80 to 1.02): it matches a Box-Cox transform with λ ≈ 0.18 of the mg/mg values, which back-transforms to the paper's mean 0.36 (SD 0.44) and median 0.20 (quartiles 0.09, 0.47) mg/mg protein. Back-transformed values are not uploaded.
Raw dataset 1 (118 lines, sex M, 12 phenotypes): each line has one pooled sample per assay, so the dataset has one row per line and the DGRPool line means equal the published values. Phenotypes:
Glycogen,
Trehalose,
Triglyceride (whole body) and
Gut_TotalBacteria,
Gut_Apersici,
Gut_Apomorum,
Gut_Efaecium,
Gut_Lplantarum,
Gut_Lbrevis,
Gut_TotalFungi,
Gut_Ascomycota,
Gut_Basidiomycota (gut qPCR). Values are copied as published, with NA where the table has NA: triglyceride is available for 88 lines,
L. brevis for 100, the other traits for 107 to 118 lines (DGRP_707 has no metabolite value). The authors' GWAS (DGRP2 web tool, with
Wolbachia status and the major inversions as covariates) used Triglyceride, Gut_Basidiomycota, Gut_Apersici, Gut_Apomorum and Gut_Lbrevis.
Checks. The uploaded values reproduce the paper: every Pearson correlation quoted in the Results (for example Basidiomycota vs triglyceride r = -0.219, P = 0.044, and
A. pomorum vs triglyceride r = -0.722) and all 27 univariate regressions of Table 1 (coefficient, SE and P, up to rounding). Eight of the 9 multivariable models of Table 1 are reproduced as well; the exception is total fungi on glycogen (0.19 (0.10), P = 0.058 from the data vs 0.09 (0.09), P = 0.29 in Table 1). The allele counts of the GWAS sheets are consistent with the number of lines with a value for each trait.
Not uploaded: the 16S rRNA and ITS amplicon sequencing of 10 lines (DGRP_026, 138, 217, 235, 354, 370, 439, 705, 837 and 900; phylum-level relative abundances shown only in Supplemental Figures 1 and 2; reads in BioProject
PRJNA1004984).