Comments from curator
Phenotyping data comes from the
Supporting Information of the paper (
journal page, also on
PMC; files from the
Europe PMC supplementary files):
-
File S2 (FileS2.xlsx): the raw data, one row per mating trial (1,322 trials), used for the raw dataset.
-
Tables S1, S2 and S3: mean ± SD of copulation latency, courtship latency and copulation duration for each of the 100 male line × female line pairings. They are not uploaded (per-pairing values cannot be stored per line), but they were used to check the raw data.
Design: 10 DGRP lines (DGRP_208, 304, 315, 360, 379, 437, 486, 517, 707 and 732) were tested in a full 10 × 10 diallel of no-choice trials: one virgin male of a line and one virgin female of the same or another line, 3-6 days old, were brought together in a chamber of a mating array and video-recorded for 1 h (25 °C, within 2 h after lights on). Each of the 100 pairings was repeated with the aim of obtaining at least 10 matings. From the videos, the authors scored the
courtship latency (from the start of the trial to the male orienting toward the female), the
copulation latency (from the onset of courtship to mounting), the
copulation duration (from mounting to separation) and whether the pair mated.
Layout of the raw dataset: each trial involves two DGRP lines, so it is uploaded twice: once under the line of the male (sex M) and once under the line of the female (sex F). The other line is given in the obsolete column
PartnerLine, and
TrialID (the row number in File S2) links the two rows of a trial. The values computed by DGRPool for sex M therefore describe the male genotype over all its female partners, and those for sex F the female genotype over all its male partners, as in the paper (means and SDs computed separately for each sex of each line). For copulation latency, which is controlled by the female, the male values measure male attractiveness and the female values female responsiveness. For courtship latency and copulation duration, which are controlled by the male, the male values measure male mating propensity and the female values female attractiveness to males. Pairings of a line with itself appear under that line once as M and once as F.
The raw dataset has:
-
CourtLat,
CopLat and
CopDur: courtship latency, copulation latency and copulation duration (s).
-
Mated: 1 if the pair mated within the hour, 0 otherwise.
- Obsolete design columns:
PartnerLine,
Age and
PartnerAge (age in days of the fly of the line and of its partner),
TestDay (test day 1-19),
Camera (1-4) and
TrialID.
The ln-transformed columns of File S2, which the authors analysed, are not uploaded: they equal the ln of the raw values.
Censored and missing values:- For the 339 pairs that did not mate within the hour, File S2 gives a copulation latency of 3,600 s minus the courtship latency (the time left until the end of the recording). The authors used these values: Table S1 is reproduced exactly only when they are included, and so is the Levene test of the paper (F
9,1312 = 2.6431, P = 0.0049, on the ln of all 1,322 values). They are kept, so CopLat is right-censored when Mated = 0 (the authors used a Cox model for it).
- File S2 gives a copulation duration of 1 s (ln = 0) to the pairs that did not mate. These were set to NA, because the paper excludes non-mating pairs from the copulation duration analyses.
- 16 pairs that mated have no copulation duration in File S2 (NA): 14 trials of test day 16 (camera 3), and 2 trials whose ln(duration) is given as 9.61 and 9.62 (about 4.1 h, which is not possible in a 1-h recording).
Checks: the means and SDs of Tables S1 (copulation latency) and S2 (courtship latency) are reproduced exactly from File S2 for all 100 pairings. Table S3 (copulation duration) is reproduced only when non-mating pairs and the 14 day-16 trials count as 1 s, so its values are not means over mated pairs. As in Figure 3, males of DGRP_208 are the most attractive (shortest mean copulation latency, 811 s) and males of DGRP_379 the least (1,609 s). Females of DGRP_208 are the most responsive (243 s), and females of DGRP_707 the least (2,092 s).
Of note:- The paper says that each pairing was repeated until at least 10 matings were obtained, with 10 to 27 trials per pairing. In File S2, pairings have 8 to 23 trials, and 25 of the 100 pairings have fewer than 10 matings (4 to 9). The proportion of trials that mated per pairing ranges from 0.17 to 1, not from 0.37 to 1 as stated. The data are uploaded as published.
- Two trials have a male aged 2 days (the paper says 3-6 days).
- The authors' per-line measures of preference were only shown in figures: female responsiveness (mean copulation latency of a female line), female choosiness (SD among the 10 pairing means of a female line) and male attractiveness (mean copulation latency of a male line). They can be computed from CopLat and PartnerLine. With 10 lines, the paper has no GWAS.