Study

Curated
Nunez et al., 2024
A cosmopolitan inversion facilitates seasonal adaptation in overwintering Drosophila
[10.1093/genetics/iyad207]

Description

Startle-response length and basal activity of F1 females from 10 DGRP lines (6 In(2L)t inverted, 4 standard) crossed to 5 deficiency stocks covering In(2L)t candidate regions (quantitative complementation). Df(2L)BSC37 failed to complement for startle response.

Comments from curator

Phenotyping data comes from the authors' GitHub repository (Cville-Seasonality-2016-2019, CODE/15.DGRP.DEFICIENCY_SCREEN), file Intermediary_data/startle.data.RDS: one value of startle-response length (sr.length) and of basal activity (basal.act.min) per F1 female, computed by the authors' scripts (4.Data.aggregation.R, 5.Phenotype.analysis.R) from their DAM2 monitor recordings. The cross codes of the file were mapped to DGRP lines with Supplementary Table 10 (crossing scheme) of the paper (iyad207_supplementary_data.zip, PMC10847723); the assay is described in Supplementary Methods 2.

Design (quantitative complementation of In(2L)t candidate regions): 5 deficiency stocks covering candidate windows of In(2L)t (balanced over CyO or SM6a) were each crossed to 5 DGRP lines homozygous for the inverted (3 lines) or standard (2 lines) arrangement: 25 crosses, 10 DGRP lines. 3-5 day-old F1 females were sorted on the curly-wing phenotype into a deficiency class (DGRP/Df) and a balancer class (DGRP/balancer), placed overnight in DAM2 monitors (20 °C, constant light, activity recorded every second) and startled at 10 AM by 5 s of vibration. Each cross was assayed on 5 consecutive days in 3 independent blocks. Activity was averaged over 90-s windows moved by 30 s.

Raw dataset (raw_1), one row per F1 female (2,247 flies, sex F), stored under the DGRP parent line. As for other F1 crosses in DGRPool, the cross is given by the name suffix _x<deficiency>, and the F1 class by _Df (carries the deficiency) or _Bal (carries the balancer); the other columns of a row are NA:
- _xBSC37: Df(2L)BSC37, dpp[EP2232]/CyO (Bloomington 7144; authors' code k2a) x DGRP 837, 32, 348 (inverted) and 853, 634 (standard)
- _xBSC455: w[1118]; Df(2L)BSC455/SM6a (Bloomington 24959; authors' code k2b) x DGRP 161, 386, 748 (inverted) and 189, 57 (standard)
- _xExel6011: w[1118]; Df(2L)Exel6011, P{w[+mC]=XP-U}Exel6011/CyO (Bloomington 7497; authors' code k5a) x DGRP 837, 32, 348 (inverted) and 853, 634 (standard)
- _xBSC109: w[1118]; Df(2L)BSC109/CyO (Bloomington 8674; authors' code k5b) x DGRP 161, 386, 748 (inverted) and 189, 57 (standard)
- _xED678: w[1118]; Df(2L)ED678/SM6a (Bloomington 8906; authors' code k9a) x DGRP 837, 32, 348 (inverted) and 853, 634 (standard)
- SRLen_... (minutes): startle-response length, the time from the fly's peak of activity after the stimulus to the first window at which its activity is back to its basal level (30-s steps; 0 when the peak is already at or below basal).
- BasAct_... (counts/min): basal activity, the mean activity per minute over the hour before the stimulus.
- Week, Day, FlyNo (obsolete design columns): experimental block (1-3), assay day within the block (1-5) and fly number of the fly ID (e.g. k2a-di4.D3_day2_week1).
About 200 flies whose activity never returned to basal were removed by the authors' script and are not in the file (no value for them). The authors' models also use the ancestral/derived haplotype of each inverted line at windows w5.2 and w9.6 (Table S10); this is a property of the lines and is not uploaded.

Not uploaded: the startle-response decay rates (Fig. 7g) are model slopes per genotype class (emmeans.data.RDS), not per line or fly; the per-fly activity time courses (modeling.data.RDS) and raw monitor files are not phenotypes. The rest of the DGRP work of the paper is a GWAS meta-analysis of 225 published DGRP phenotypes (Supplementary Tables 8-9), i.e. reused data. Model outputs of the population genomics are on Zenodo (10.5281/zenodo.7305042).

Mismatch with the paper: refitting the authors' test (lmer(value ~ genotype x F1 class + (1|block)) vs the additive model, likelihood-ratio test) on this file gives chi-square values close to Supplementary Table 11 for basal activity (e.g. 0.37 vs 0.37 for Df(2L)BSC455, 2.44 vs 2.42 for Df(2L)ED678) but not for startle-response length (e.g. 2.42 vs 3.50 for Df(2L)BSC37, P = 0.12 vs 0.061). The GitHub file is probably from another run of the authors' pipeline; it is the only per-fly file available.
Back to studies