Comments from curator
Phenotyping data comes from the figure source data of the paper (
Li et al. 2023, eLife 12:e83856), available in the
Figures and data section:
-
Figure 1—source data 1: sheet "Figure_1C" (rolling latency of each larva) is used for raw dataset 1; sheets "Figure_1D" (rolling probabilities) and "Figure_1E and F" (the four metrics used for the GWAS, identical to
Figure 1—figure supplement 1—source data 1) are used for the summary dataset.
-
Figure 2—figure supplement 2—source data 1: sheet "Figure 2 - figure supplement 2B" (
bero RT-PCR quantification) is used for raw dataset 2.
Assay: Heat Probe Assay on wandering third-instar larvae of 38 DGRP lines, raised at 25 °C (12 h light/dark, standard food). Larvae of both sexes were tested without sexing, so the sex is NA. Each larva was touched laterally on abdominal segments A4-A6 with a heat probe at 46 °C, and the rolling latency is the time from the stimulation to the initiation of the first 360° roll. The authors coded larvae that did not roll within 10 s as 10.05 s. The paper reports 30 or 31 larvae per line, tested on several days.
Raw dataset 1: one row per larva, with its rolling latency in s (RollLatency), copied unchanged from the sheet (including the float noise of the stored values, e.g. 4.800000000000001). The 10.05 s code was kept, as the authors used it in Figure 1C, in their median latency and in their statistical tests. Therefore the mean computed by DGRPool includes the non-rolling larvae as 10.05 s and is not the authors' "average latency" (see the summary dataset). The share of larvae with a value below 10.05 is the rolling probability within 10 s.
One larva per line is missing from the per-larva sheet: it holds 29 larvae for 33 lines and 30 for 5 lines (1107 larvae), one fewer than the 30 or 31 larvae per line of the figure legend and of the authors' per-line metrics. For 32 lines, adding a single larva (a non-rolling larva at 10.05 s for 18 of them) reproduces all five authors' metrics exactly. For DGRP_365, DGRP_391, DGRP_427, DGRP_712 and DGRP_859, one of the three rolling probabilities still differs. DGRP_399 has 30 larvae both in the sheet and in the metrics, but one larva differs. As a result, the medians computed by DGRPool from raw dataset 1 differ from the authors' medians by up to 0.89 s (identical to 2 decimals for 6 of the 38 lines), which is why the authors' values are also provided in the summary dataset.
Summary dataset: the authors' per-line metrics, copied unchanged:
- mn_RollProb2s, mn_RollProb5s and mn_RollProb10s: rolling probability within 2, 5 and 10 s, i.e. the percentage of the larvae of the line that started rolling within that time (Figure 1D).
- mn_RollLatRollers: the authors' "average latency". It is the mean latency of the larvae that rolled within 10 s, the larvae coded 10.05 s being excluded. The paper does not define it, but this definition reproduces the authors' value exactly for 21 of the 38 lines (the 20 lines whose missing larva did not roll, and DGRP_517), whereas the mean including the 10.05 s values differs from it by up to 3.01 s.
- md_RollLatencyAuth: the authors' median latency, including the larvae coded 10.05 s.
The probabilities within 5 and 10 s, the average latency and the median latency are the four metrics that were submitted to the DGRP2 web-based GWAS pipeline (Figure 1F for the median, Figure 1—figure supplement 1B for the other three). The probability within 2 s was not mapped.
Two values of the source are inconsistent and were kept as published: DGRP_427 has a rolling probability within 2 s of 15.38%, which is impossible with 30 larvae (the per-larva data give 4 of 30, i.e. 13.33%), and DGRP_391 has 0% within 2 s although one of its larvae in the sheet rolled at 1.27 s.
Raw dataset 2:
bero (
CG9336) mRNA level in the larval central nervous system of 4 DGRP lines chosen for their extreme rolling latency: DGRP_208 and DGRP_315 (fast, minor allele C of the
bero-associated SNP 2L_20859945_SNP) and DGRP_391 and DGRP_705 (slow, major allele T). It was measured by semi-quantitative RT-PCR as the background-subtracted band intensity of
bero divided by that of
αTub84B on one gel, with one pooled sample of the CNS of two wandering third-instar larvae per line. There is one row per sample (bero_mRNA_CNS, a unitless ratio), and there are no replicates. Values were copied unchanged.
Not uploaded: the
w1118 and Canton-S controls (Figure 1B), Canton-S and the
bero knockout (Figure 2—figure supplement 2B), and all other experiments, which used GAL4/UAS, mutant or RNAi stocks, not DGRP lines. The 2L_20859945_SNP alleles listed per line in Figure 2—figure supplement 2A are DGRP genotypes, not phenotypes. The paper's preprint is
bioRxiv 10.1101/2022.10.20.513010.