Comments from curator
Phenotyping data comes from the authors'
Zenodo deposit (Data Availability Statement of the
paper; preprint
bioRxiv 2024.05.09.593162):
-
flypad.csv: FlyPAD two-choice feeding assay, one row per fly (1,854 mated females of 122 DGRP lines), used for the raw dataset 1.
-
survival96.csv: survival after oral infection, one row per fly (843 flies of 20 DGRP lines), used for the raw dataset 2.
The per-line files of the deposit (flypad_preference_summary.csv, total.sips_summary.csv, Pref-Sips-correlation.csv) are reproduced exactly by the means of raw dataset 1 (same number of flies, mean, SD and SE for every line), so no summary dataset was added. The preference index files lack DGRP_502 (18 flies), which is in flypad.csv and among the 122 lines of the paper.
FlyPAD assay (raw dataset 1, sex F): 2-6-day-old mated females were wet-starved for 18-22 h, then placed individually in a FlyPAD arena with two 1 µL substrates: a clean food (LB broth, 2% agar, 5% sucrose) and the same food with an overnight culture of
Pseudomonas entomophila (strain L48, OD
600 = 1). Each contact of the fly with a food is recorded as a sip, over 30 min.
-
SipsClean_Pe,
SipsInfected_Pe and
SipsTotal_Pe: cumulative number of sips on the clean food, on the contaminated food and on both (Figure 1B; unit counts/30min).
-
PrefIndex_Pe: preference index, (sips clean - sips contaminated) / sips total, from -1 (only the contaminated food) to 1 (only the clean food); positive values mean avoidance of the pathogen. The authors' GWAS used its line means (run on DGRPool).
- Design columns kept as obsolete phenotypes:
Replicate (number of the fly within its line, 1-24; the "replicate (individual fly)" random effect of the authors' model),
Day (experimental day, 1-24),
TimeSlot (block of the day, A-D; missing for 2 flies) and
FoodFirst (substrate pipetted first: I contaminated, NI clean).
All values are copied unchanged, including the 47 flies with fewer than 10 sips in total, which the authors kept.
Survival after oral infection (raw dataset 2): the 10 lines with the highest and the 10 lines with the lowest preference for the contaminated food were wet-starved for 6-8 h, exposed for 24 h in cohorts of 10 flies to
P. entomophila (100 µL at OD
600 = 50 in 5% sucrose on a filter disc), then moved to standard vials.
DeathTime_PeOral is the time (hours post infection) at which the fly was found dead (checks at 6, 16, 24, 32, 42, 48, 60, 72 and 96 h). The cohort (vial letter A-G within the line) and the infection day (A or B) are kept as the obsolete design phenotypes
Vial and
InfectionDay. The paper does not give the sex of these flies (they were "reared and treated as above", so probably mated females as in the FlyPAD assay): their sex is NA.
Of note:
- The survival file has only the infected flies (not the control cohorts fed sucrose), and all 843 flies are coded as dead (censor = 1), including the 8 flies at 96 h, as in the authors' Cox model, although the legend of Figure 3A mentions flies censored alive at 96 h. It holds 843 of the 1,400 planned flies (7 cohorts of 10 per line): from 2 flies (DGRP_822) and 13 (DGRP_309) to 70 (DGRP_101, DGRP_380) per line, so some line means rest on few flies. A Cox model of these data reproduces the paper's hazard ratio between the two groups of lines (1.099, SE 0.07, P = 0.18).
- The group with the highest preference for the pathogen includes DGRP_426 (11th lowest mean preference index) instead of DGRP_853 (10th, tested on 3 flies only). The legend of Figure 3A says five lines per group, the methods ten; the file has ten.
- The paper says each line was tested 16 to 24 times, but flypad.csv has 3 to 24 flies per line (fewer than 16 for 65 lines).
- The lines are the authors' stocks, cleared of
Wolbachia by tetracycline treatment, whereas DGRPool's GWAS uses the
Wolbachia status of the DGRP annotation as a covariate (as the paper notes).
The data reproduce the paper: line means of the preference index from -0.664 ± 0.064 SE (DGRP_879) to 0.225 ± 0.161 (DGRP_380), negative for 108 of the 122 lines, mean -0.22 and median -0.31 over all flies (Table 1); line means of total sips from 35.5 ± 13.8 (DGRP_790) to 1,047.1 ± 140.6 (DGRP_059); correlation between the line means of total sips and preference index r = -0.20, P = 0.029 (Figure 1C).
Not uploaded: the FlyPAD data of the DGRP-derived outbred AOx population (other pathogens and starvation periods; not per line), the FlyPAD data of the IMD pathway mutants (
w1118,
Relish,
PGRP-LC,
PGRP-LB; not DGRP lines), and two files of data from other studies: oral_corr.csv (percentage of flies dead 3 days after oral
P. entomophila infection, from Bou Sleiman et al. 2015,
DGRPool study 4) and dietcorr.csv (pupation and eclosion on high-sugar and high-protein diets, from Havula et al. 2022,
study 148).