Comments from curator
Phenotyping data comes from the
Supplementary material of the bioRxiv preprint (
bioRxiv, v1 of 2026-07-01;
supplementary material):
-
Supplementary Dataset 1 (media-3.zip, "Supplementary Dataset 1.txt"): raw data of each mated DGRP female (2,084 females of 68 lines): clock times of introduction, start and end of mating and of plug ejection, ejection and observation times, time to mating and mating duration. Used for the raw dataset.
-
Supplementary Dataset 2 (media-4.zip, "Supplementary Dataset 2.xlsx"): one row per experiment (one line on one day; 74 experiments): pairs set up and mated, proportions mated and ejected, the survival-model median ejection time (GWAS phenotype), the raw median, and the GWAS covariates. Used for the summary dataset.
Design. Virgin females of 68 DGRP lines, aged 3-5 days, were each paired with a virgin Canton-S male (the same male stock in all experiments, to study female variation; the values may include female responses to this male genotype). The times of introduction and of the start and end of copulation were recorded. After copulation, the female was moved to an individual 3D-printed observation chamber and checked every 10 min under blue light for the autofluorescent mating plug; the ejection time is the time from the end of mating to the first check without the plug. About 50 pairs were set up per experiment (10-70; 3,086 in all, the paper's "over 3000 matings"), of which 4-46 mated and were observed. Six lines (DGRP_307, DGRP_357, DGRP_439, DGRP_732, DGRP_738, DGRP_852) were assayed on two days. Experiments ran from 2019 to 2022 (four researchers). All values are those of the DGRP females (sex F).
Raw dataset (one row per mated female, 2,084 rows; values as published in Dataset 1):
-
PlugEjectTime: time from the end of mating to plug ejection (min); NA for the 267 females (12.8%) that had not ejected by their last observation.
-
PlugEjected: 1 if the female ejected the plug, 0 if not; its line mean is the proportion that ejected (Figure 1B).
-
PlugObsTime: time from the end of mating to ejection, or to the last observation for the females that did not eject (right-censored; min).
-
MatingLatency (h) and
MatingDuration (min): time from introduction to the start of copulation, and copulation duration, from the recorded clock times. They are in Dataset 1 but not analysed in the paper.
-
ExptDate,
Researcher and
FemaleID (obsolete): design columns (day of the experiment, researcher code, pair number of the day).
Censored values. The authors treated the females that did not eject as right-censored at their last observation, in a survival model. So PlugEjectTime is NA for them (its line mean covers the ejecting females only, and is biased downwards for lines with many non-ejectors), while PlugObsTime and PlugEjected hold the censored data the authors analysed. The paper describes a five-hour observation window, but the females were observed until the end of each experiment day: ejections were recorded up to 491 min after mating (99 females ejected after 5 h and count as ejected in the authors' data), and the censoring times of the non-ejectors range from 47 to 496 min (104 observed less than 5 h; for 58 non-ejectors of 9 experiments of 2022, the authors' "5 hr rule" set it to the last observation of the day).
Summary dataset (authors' values per experiment, from Dataset 2):
-
md_PlugEjSurv: median ejection time (h) estimated by a log-normal parametric survival regression (R survival::survreg) on the females of the experiment, non-ejectors censored. Its log is the authors' GWAS phenotype (lme4qtl::relmatLmer, with time, the four major inversions and
Wolbachia as fixed effects, and line, genetic relatedness and researcher as random effects; DGRP2 genotypes, 1,837,106 SNPs).
-
mn_MatedProp: proportion of the pairs set up that mated (frac.mated, 2 decimals; not analysed in the paper).
There is one row per line. For the 6 lines assayed on two days, the first experiment is in md_PlugEjSurv and mn_MatedProp and the second in
md_PlugEjSurv_rep2 and
mn_MatedProp_rep2 (the GWAS used both). DGRP_365, DGRP_399 and DGRP_714 have a single experiment, labelled "second" in Dataset 2; it is in the main phenotypes. Refitting the survival model on the raw dataset reproduces the 74 medians (max difference 2e-9 h). The proportions ejected and the raw medians of Dataset 2 (frac.ejected, raw.med = median of PlugObsTime) are reproduced from the raw dataset for all 74 experiments; DGRPool's line values equal them for the 62 lines assayed once, and pool the two days for the 6 others. The DGRP inversion genotypes and
Wolbachia status (DGRP annotations), the GWAS time covariate and two undocumented columns (orig.pheno, chk1) were not uploaded.
Differences with the paper.- The text and Figure 1 say 69 DGRP lines; both datasets have 68 lines (74 experiments).
- The paper gives the average of the line medians as 2.81 ± 1.5 h; the 74 medians of Dataset 2 have mean 2.81 h and SD 1.25 h.
- Repeatability: the paper reports r = 0.986 (t = 11.741, df = 4, P = 0.0003) and Spearman rho = 1 for the 6 lines assayed twice. These values are obtained by pairing the smaller with the larger median of each line (as plotted in Supplementary Figure 7); pairing the first with the second experiment gives r = 0.964 (P = 0.002) and rho = 0.886.
The preprint also compares germline-less females (daughters of
tudor mothers) with control females, mated to Canton-S males (Figure 3, Supplementary Dataset 5); they are not DGRP lines and were not uploaded. The GWAS results are in Supplementary Datasets 3 and 4.