Comments from curator
Phenotyping data comes from the
Supplemental Material of the paper (
journal page, also on
PMC, section Supplementary Material):
-
Supplemental Table S1 (supp_gr.243030.118_Supplemental_Table_S1.xlsx): the 453 metabolites measured in each of the 240 samples (40 DGRP lines x 2 sexes x 3 replicates), used for the raw dataset.
-
Supplemental Table S8 (supp_gr.243030.118_Supplemental_Table_S8.xlsx): line means of body weight, free glucose, glycogen, free glycerol, triglyceride, total protein, thorax length and thorax width, per sex, used for the summary dataset (no per-replicate values were published for these traits).
Design. The 40 DGRP lines are minimally related, have the standard karyotype for all common inversions and are free of
Wolbachia. Flies were reared on cornmeal-molasses-yeast medium at 25°C (12 h light-dark). For the metabolome, three replicates of 100 flies aged 3-7 days were collected per line and sex and flash frozen, and the 240 samples were profiled by Metabolon, Inc. (UPLC-MS/MS on four platforms: LC/MS positive early and late, negative and polar).
Raw dataset (metabolome). One row per pooled sample of 100 flies (240 rows), one phenotype per metabolite (453), plus the replicate number of the sample (1-3, obsolete design column). Values are the authors' "scaled measurement" (peak area normalised in run-day blocks to a median of 1; relative abundance, no unit), uploaded exactly as in Table S1. Metabolite names are abbreviated in the phenotype names (e.g. PC_16_0_18_2 for 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2), LPC_ for lyso-GPC, DAG_/MAG_ for di/monoacylglycerols, AcetylAla for N-acetylalanine, AlaLeu for the dipeptide alanylleucine); each phenotype description gives the full name (pathways and Metabolon, KEGG and PubChem identifiers: Supplemental Table S1 of the paper). 53 metabolites were not confirmed with a standard (marked * in the paper; their descriptions say so).
Of note: Table S1 has no missing value. In 187 metabolites, three or more samples (8387 values in all) have exactly the minimum value of the compound: these are probably non-detects imputed with the minimum (Metabolon's usual practice; the paper does not say). The authors used these values in their analyses, so they are kept as published. In particular, four metabolites were detected only in males, so all their female values are the same minimum (PC_P16_0_18_2, PC_P16_0_16_1, Glycerol2P, Mevalonate5P), and one only in females (LPI_18_0, constant in males); the authors' per-sex ANOVAs give them F = 0.
Summary dataset (organismal traits, Table S8). Body weight was measured on groups of 10 virgin flies per tube (the paper says 10 replicates per line and sex), so
mn_Weight10Flies is the weight of 10 flies (mg), not of one fly. The same flies were homogenised in 250 µL PBS, and free glucose, glycogen, free glycerol, triglyceride and total protein are concentrations in that homogenate, in ng/µL as published (= µg/mL). Thorax length and width are in mm; they were not done for DGRP_229 (NA).
Reproduction of the paper. The per-sex means of each metabolite computed from the raw dataset equal the "Mean of Response" of Supplemental Table S3 (453/453 metabolites, both sexes), and one-way ANOVAs per sex reproduce its F values and variance components; 371 (females) and 355 (males) metabolites have FDR < 0.05 in Table S3, as in the text. The Spearman correlations between the Table S8 traits (Table S10) and between the metabolite line means and the Table S8 traits (Table S9, 164 pairs per sex) are reproduced from the uploaded data.
Not uploaded: the module principal components (not published per line), the mQTL GWAS results (Tables S5, S6), the trait correlations and network tables (Tables S9-S12) and the RNA-seq of 39 of the 40 lines (
Everett et al. 2020, GEO GSE117850). The metabolite-trait analyses also used starvation resistance, startle response, waking activity, virgin lifespan and male aggression taken from earlier papers (Harbison et al. 2004 and 2013, Jumbo-Lucioni et al. 2010, Huang et al. 2012, Ivanov et al. 2015, Shorter et al. 2015; see DGRPool studies
15,
18,
21 and
30); they were not measured in this study and are not uploaded.