Comments from curator
Phenotyping data come from the authors' figshare dataset
doi:10.6084/m9.figshare.12937484.v1 ("Reaction norm lifespan and fecundity data", CC BY 4.0), files
reaction.norm.datalongformat.csv (lifespan) and
fecundity.reaction.norm.csv (egg laying), and from
Tables S2-S6 and S8-S12 of the
Supplementary information (jexbio-223-230185-s1.pdf).
Assay: mated females of 5 DGRP lines (DGRP_195, 217, 239, 362 and 853, a subset of the 11 lines of McCracken et al. 2020, Science Advances, DGRPool study 94) were reared and mated for 48 h on 8% yeast food and kept in demography cages of 70-125 females (mode ~100) on 8% food until age 3-4 days. From then on each cage received one of five diets for life: 0.5, 2, 5, 8 or 14% autolysed yeast (13% sugar, 6% cornmeal, 1% agar, 0.225% nipagin; the cornmeal was halved in the 14% food). A second opening of each cage held either a vial of 2% water-agar (
water supplementation) or an empty vial (
control), so that the flies could drink without eating. There were 5 cages per line, diet and water treatment (50 per line), plus 2 extra 2%-yeast cages of each water treatment for DGRP_195: 254 cages and 25,519 females in total, as stated in the paper. A census was taken every 48 h: dead females were counted and removed and the food vial was replaced. For egg laying, the feeding vials of the census at age 11 or 12 days were imaged and the eggs counted with QuantiFly image analysis (relative estimates, comparable between cages).
Phenotype names: the diet is Y0p5, Y2, Y5, Y8 or Y14 (0.5 to 14% yeast) and the water treatment is Ctrl (empty vial) or Water (water-agar vial).
Raw dataset 1 (lifespan, one row per female, sex F): the source file has one row per female (columns Genotype, Cage, Age, At.Risk, Treatment, Transfer, Dead, Water). The 10 diet x water conditions are independent cages, so they are stacked compactly: row k of a line holds the k-th female of each condition (ordered by cage, then age).
-
Lifespan_<diet>_<water>: age in days (Age column) at the census at which the female was found dead (Dead = 1).
-
CensorAge_* (obsolete): age at the census at which the female was right-censored (Dead = 0; 2,398 females, 9.4%; the paper does not say why they were censored). Their Lifespan is NA, so the line means computed by DGRPool are means of the observed ages at death and ignore the censored females, whereas the authors' Cox models treat them as right-censored.
-
Cage_* (obsolete): the authors' cage number (numbered within each line; cage was a random effect in the authors' models).
Not uploaded: At.Risk (the number of females of the cage still alive at that age, i.e. derivable from the rows) and Transfer (a cage-level code 1-10 that the paper does not describe or use in its analyses).
Raw dataset 2 (egg laying, one row per cage, sex F): 233 of the 254 cages have a fecundity record (missing: DGRP_195 cages 51, 52, 53 and 56, which are 2%-yeast cages; DGRP_217 cages 21, 48, 49 and 50; DGRP_239 cages 14-19, 21, 22, 24 and 25; DGRP_853 cages 1, 2 and 3). Conditions are stacked compactly as above.
-
Eggs_*: QuantiFly egg count of the cage's feeding vial at age 11-12 days (column Corrected.count., which the authors used for their outputs; the "raw egg counts" of Fig. S2A, analysed as log10 in Table S13). The values are calibrated estimates, not integers.
-
EggsFem_*: the same count divided by the number of females alive in the cage at that census (column AEggs; the "mortality-corrected counts" of Fig. S2B, analysed as log10 in Table S14).
-
NFem_*,
EggAge_* and
FecCage_* (obsolete): the number of females alive in the cage at that census (At.Risk; it equals the count in the lifespan file), the age of the females at the census (11 or 12 days), and the cage number (same numbering as Cage_* of raw dataset 1).
Not uploaded: Predicted.count. (the uncalibrated QuantiFly prediction; Corrected.count. = 1.2956 x Predicted.count. - 13.59 in every row), CI, and the Dead and Censor columns. The lifetime reproductive fitness estimates of Fig. S3 (mortality-corrected counts multiplied by the restricted mean lifespan) are given only in that figure.
Summary dataset (sex F): per-line log hazard ratios of mortality from the authors' within-line mixed Cox models (cage as random effect), as printed in the "estimate" columns of the tables (summary type hazard ratio; the hazard ratio is exp of the value, and positive values mean higher mortality, i.e. shorter life):
-
hr_log_<diet>vs8_Ctrl (Tables S2-S6, rows "0.5% yeast" to "14% yeast"): each diet vs. 8% yeast in control cages (the dietary reaction norms of Fig. 2A).
-
hr_log_Water_Y8 (Tables S2-S6, row "water supplementation"): water vs. control on 8% yeast.
hr_log_Water_Y0p5,
_Y2,
_Y5 and
_Y14 (Tables S2-S6, columns "Effect of water, compared to no water"): water vs. control on each other diet, i.e. "water supplementation" + "<diet> * water" (Fig. 2B).
-
hr_log_<diet>xWater (Tables S2-S6, rows "<diet> * water"): the interaction terms of these models.
-
hr_log_<diet>vs8_Water (Tables S8-S12): each diet vs. 8% yeast in a model fitted on the water-supplemented cages only (Fig. S1).
Standard errors and p-values were not uploaded. The models fitted across the 5 lines with DGRP_195 as reference (Tables S1 and S7) and the fecundity models (Tables S13 and S14) were not uploaded.
Checks: refitting the within-line Cox models on the figshare lifespan file (survival::coxph with a log-normal frailty on cage) gives all 85 published estimates within 0.004. Refitting the linear models of Tables S13 and S14 on the fecundity file gives all their published estimates exactly. The paper reports no per-line mean or median lifespan (only survival curves, Fig. S4), so the line means computed by DGRPool cannot be compared with published values.
