Comments from curator
Phenotyping data comes from the authors'
Dryad deposit (
10.5061/dryad.6wwpzgmv8, mirrored on
Zenodo), file
InbredOutbred_DGRP.tar.gz: the raw DAM monitor files, the per-minute activity and sleep of each fly (ProcessedData), the per-genotype summaries (SummaryData) and the authors' R scripts. In the paper, these DGRP data are only shown as a figure (
Supplementary Figure S3 (contd), sleep time) and compared with Harbison et al. 2013 in the Results (section 3.1).
Context. The main study maps sleep QTL in 787 F1 genotypes from crosses between DSPR recombinant inbred lines (DSPRF1; Supplementary Table S1), which are not DGRP lines and are not uploaded. Because the inbred DGRP lines of Harbison et al. 2013 slept much more in the light phase than the outbred DSPRF1 flies, the authors tested inbreeding directly: they measured sleep in
24 inbred DGRP lines and in 22 F1 genotypes from crosses between pairs of these lines (female line x male line). Only the 24 inbred lines are uploaded: each F1 is a hybrid of two different DGRP lines (there is no common tester), so it cannot be stored under one DGRP line. The authors found no clear difference between inbred and F1 flies (the F1 slept less than both parents for 10 of the 22 crosses in the light phase and 3 in the dark phase). The 168-line DGRP histogram of Supplementary Figure S2 is taken from Harbison et al. 2013 (
DGRPool study 15) and is not new data.
Design. Virgin females were collected under CO
2, kept in groups of 8-12 for 3-4 days, then loaded singly without anesthesia into DAM2 monitor tubes (TriKinetics) with molasses food (Nutri-Fly) on 21 April 2017; 8 flies per genotype, in one quadrant of a monitor. Flies were reared and assayed at 25 °C, 50% relative humidity and 12 h light : 12 h dark (lights on at 08:00). Activity (beam crossings) was recorded every minute, and
three 24 h periods are analysed, from lights on (08:00) on 23 April to 26 April 2017 (the methods of the paper give four periods, which applies to the DSPRF1 panel; the authors' script notes that this experiment ran for 3 days). A fly is asleep in any fixed 5-min window without a beam crossing. Lines were given as Bloomington stock numbers and mapped to DGRP lines (the authors' own conversion table agrees).
Raw dataset 1 (192 flies of 24 lines, females): one row per fly. The values were computed by us from the authors' per-minute ProcessedData files, with the definitions of their R script (Code01_AnalysisCodeAndNotes.R):
-
LightSleep,
DarkSleep: minutes asleep in the 12 h light (dark) phase, mean of the 3 days;
-
LightBoutN,
DarkBoutN: number of sleep bouts in the light (dark) phase per day, mean of the 3 days (a bout is cut at the light transitions; a phase without any bout counts as 0);
-
LightBoutLen,
DarkBoutLen: mean length (minutes) of all the fly's sleep bouts in the 3 light (dark) phases;
-
LightWakeAct,
DarkWakeAct,
TotalWakeAct: beam crossings per awake minute in the light phases, the dark phases, or the whole recording.
Monitor and
Tube are kept as obsolete design columns. The authors' script sets the activity of a fly to NA from 12 h before its death and removes tubes with fewer than 100 beam crossings, then drops every fly with any NA. These 9 flies (one each of DGRP_324, 335, 360, 380, 486, 714 and 732, two of DGRP_786) are kept as rows with NA, so there are 6 to 8 flies per line.
Summary dataset (24 lines, females). The DGRPool line means of LightSleep and DarkSleep equal the authors' genotype means (SummaryData) exactly for the 24 lines. For the other traits, the authors' genotype means pool all flies (bout length over all the bouts of the genotype, waking activity over all its awake minutes), and their bout-number code skips the phases without any sleep bout (7 light phases in 5 flies of DGRP_208, 301, 799 and 852, and 9 dark phases in 5 flies of DGRP_365, counted as 0 here). Their means therefore differ from the DGRPool means (r = 0.94 to 0.999), and they are uploaded as published in the SummaryData files:
mn_LightBoutNAuth,
mn_DarkBoutNAuth,
mn_LightBoutLenAuth,
mn_DarkBoutLenAuth,
mn_LightWakeActAuth,
mn_DarkWakeActAuth and
mn_TotalWakeActAuth. They are computed like the DSPRF1 genotype means of Supplementary Table S1. Recomputing the authors' values from ProcessedData with their method reproduces SummaryData exactly for all 46 genotypes and 9 traits.
Use in the paper. Only light- and dark-phase sleep time of the DGRP lines are used. The averages of the 24 line means are 223 min (light) and 404 min (dark), and they correlate with the female line means of Harbison et al. 2013 at r = 0.74 (light) and 0.72 (dark), as computed by the authors' script Code02_AcrossLabCorrelations.R; the text of the paper swaps the two values ("r = 0.72 for light-phase sleep time and r = 0.74 for dark-phase sleep time"). The averages of Harbison et al. for these lines are 377 and 558 min, as in the paper. Bout number, bout length and waking activity were measured on the same flies, but the paper does not report them for the DGRP lines (only for the DSPRF1); total waking activity is not used anywhere in the paper.
Of note: the authors' script Code02_AcrossLabCorrelations.R says that the male line means of Harbison et al. 2013 were used for the comparison, but the file it reads holds the female line means (equal to the female values of DGRPool study 15 for all 168 lines), as the paper says. Not uploaded: the DSPRF1 and DSPR RIL data, the head eQTL analysis, the RNAi knockdowns of
Ddc,
dysc and
tim (
elav-GAL4 > UAS-RNAi flies, not DGRP lines) and the overlap analysis of the DGRP GWAS hits of Harbison et al. 2013.