Comments from curator
Phenotyping data comes from the authors' Zenodo record
10.5281/zenodo.5784716 (also on the
de Bivort lab companion page), file
handMetaData.2021.12.01.mat, MATLAB table
allHandMetaData: one row per fly for all 183,496 flies (569 genotypes) of the de Bivort lab Y-maze experiments compiled for the paper, with their behavioral measures and metadata (Table 1 of the paper). The paper's data availability statement says the data are available from the authors; the Zenodo record is the one cited in its Methods. The table is a MATLAB object; its decoded values reproduce the paper's numbers (183,496 flies, 569 genotypes, 79.8 M turns, mean turn bias 0.496, r = 0.0357 between bias magnitude and number of turns, 26,175 flies on potato media, 223 DGRP genotypes with 64,838 flies).
What was uploaded: the DGRP flies that are not already in DGRPool (study 160), in one raw dataset with one row per fly (44,292 flies, 88 DGRP lines; 36,033 females, 6,388 males and 1,871 flies of mixed, unrecorded sex, uploaded with sex NA):
-
The serotonin pharmacology experiment of the paper (Fig. 5B; experimenter Erika Gajda, Nov 2015 - Jun 2016, 35,241 flies of 86 DGRP lines; the paper's n = 38,316 flies also counts 3,075 Canton-S flies, not uploaded). Flies were reared from egg-laying on media with alpha-methyl-tryptophan (aMW, a serotonin synthesis inhibitor) or 5-hydroxytryptophan (5-HTP, a serotonin precursor), or on the matching control media, with the number of parents adjusted line by line to target the same number of progeny per vial, from per-line progeny counts (the paper cites them as Akhund-Zade et al. 2020, whose reference is a thermal-preference preprint; they are probably those of Akhund-Zade et al. 2021, DGRPool study 55). Three drug/control pairs: 20 mM aMW on potato media vs potato media with ascorbic acid (PotMW20 vs PotAA, females), 50 mM 5-HTP on potato media vs potato media (PotHTP50 vs Pot, females), and 50 mM 5-HTP on agar media vs agar media (AgHTP50 vs Agar, females and males); plus a dose series on DGRP_181 and DGRP_441 (PotMW10, PotMW25, PotMW50, PotHTP10, PotHTP25).
-
Other experiments on DGRP_045, DGRP_105, DGRP_535 and DGRP_796 (9,051 flies), also part of the compiled data set: a 2015 pilot of the drug experiment (Jamilla Akhund-Zade; 5-HTP and aMW in cornmeal-dextrose or agar media, with untreated and ascorbic-acid controls: CM, CMAA, CMHTP10, CMMW10, Agar, AgHTP10, AgHTP25, AgHTP50, AgMW), the 2013 environmental-enrichment experiment of
Akhund-Zade et al. 2019 (J Exp Biol 222:jeb202234; Sandra Ho; CM, EnrMild, EnrHigh, females and males), mixed-sex flies of October 2013 (Chelsea O'Leary; CM), 2015 controls of thermogenetic experiments tested at 22 and 32 °C, including F1 of DGRP_045 and DGRP_535 crossed with a stock labelled 'shi' (presumably UAS-shibire[ts]) (Kyobi Skutt-Kakaria; CM, CM32C, CM_xShi), and 2021 flies aged 1-16 days tracked with MARGO (Ryan Maloney; CM).
Phenotypes: three behavioral measures per fly, as provided (full precision):
TurnBias (fraction of right turns),
NumTurns (number of turns in the assay, default length 2 h) and
Switchiness (left-right alternations relative to those expected from the fly's turn bias; 1 = random sequence). Each phenotype name ends with the rearing condition (the authors' 'expCond' label), so a row has values only in the three columns of its condition:
Pot,
PotAA,
PotMW20,
PotHTP50,
Agar,
AgHTP50,
PotMW10,
PotMW25,
PotMW50,
PotHTP10,
PotHTP25,
AgHTP10,
AgHTP25,
AgMW,
CM,
CMAA,
CMHTP10,
CMMW10,
CM32C,
CM_xShi,
EnrMild,
EnrHigh. The descriptions of the phenotypes give each condition. The F1 flies with the 'shi' stock are stored under their DGRP parent line, with the suffix _xShi. The design columns FlyID, Experimenter, ExpDate, ExpTemp, Age, Tracker (acquisition software), Box, Tray and Maze are kept as obsolete phenotypes. All flies are kept, as in the authors' analyses, including 90 flies with fewer than 50 turns (mostly autotracker, MARGO and 2015 thermogenetic-control data; their turn bias is imprecise) and 2 flies with a turn bias but no number of turns. Flies reared on potato media (with or without drug) include a second mode with a negative relation between switchiness and number of turns (Fig. 4B).
Note on the CM and Agar columns: like the authors' condition labels, CM pools the untreated flies of five experiments (2013-2021) on the four lines above, and Agar pools the 2015-2016 screen with the 2015 pilot for DGRP_045, 105, 535 and 796 (the authors' Fig. 5B code also pools them). Use the Experimenter and ExpDate columns to separate them.
The paper's trait is variability. The authors measured the variability of each genotype in each condition as the coefficient of variation (SD/mean over flies, sexes pooled) and compared drug and control media with paired t-tests across genotypes (Fig. 5B; script handMetaGxEVarAnalysis.m). DGRPool computes the same CV for each line, separately per sex. 5-HTP increased the variability of the number of turns by 16% (potato) and 25% (agar); the other effects were small (paper: -10% to +7%). From the uploaded flies (sexes pooled, as the authors), the change of the mean CV is +15.5% (39 lines) and +24.3% (29 lines) for the number of turns, and between -10.6% and +11.1% for the other pairs. The paper states n = 157 genotypes for Fig. 5B; the file has 86 DGRP lines (88 with the pilot lines) in these conditions. There is no GWAS in this paper.
Not uploaded:- The flies of Ayroles et al. 2015 (PNAS 112:6706; experimenter 'chelsea', Jul 2012 - Jan 2013: 20,546 flies, the DGRP screen, the re-measured parental lines and the F1 between DGRP lines): these data are DGRPool study 160, uploaded there from the authors' earlier file (18,811 of its 18,814 screen flies and all 365 parental flies match a fly of this file by line, number of turns and turn bias).
- 118,658 flies of the 346 non-DGRP genotypes (Canton-S, Gal4 and UAS lines and other laboratory stocks), including the Canton-S controls of the drug experiment.
- Turn clumpiness, which the authors excluded because its formula changed during data collection (its scale differs by software, e.g. median 0.0006 with autotracker vs 0.22 with ymaze31 in the drug experiment); the Levene-transformed values (lev_handedness, lev_numTurns, lev_switchiness), derived values for the authors' linear models of variability; eyeColor (white locus: wild type for every DGRP fly, wild type plus mini-white for the _xShi F1), arrayFormat and the analysis software.
- The moment-resampling results (momentEstimates) and the per-experimenter tables of the same file, which are subsets of allHandMetaData.
Inconsistencies in the source: the Table 1 notes of the labels 'amwagar' ("control agar media supplemented with 15 mg/mL ascorbic acid") and 'ctrlaanormal' ("control potato media") do not match the labels or the experimental design (amwagar flies were tested alongside agar controls; ctrlaanormal flies alongside untreated cornmeal and 5-HTP/aMW cornmeal flies) and look shifted by one row: they are described here as aMW in agar media and ascorbic-acid cornmeal control. The ascorbic acid dose is "15 mg/mL" in Table 1 and 15 mg per 60-mL vial in the Methods, which also say it was added to the 5-HTP groups and their controls, while the authors' code pairs the ascorbic-acid potato control with aMW. The date column mixes YYYYMMDD and YYMMDD formats (converted). The paper's Fig. 5A legend gives '3466' non-DGRP genotypes; the file has 346.