Comments from curator
Phenotyping data comes from
Supplemental Table S1 of the bioRxiv preprint (sheet "table.S1" of media-2.xlsx, "Supplemental tables S1-8, S10",
bioRxiv, Supplementary material): for each of the 46 DGRP lines tested, the mean percentage of pale R7 photoreceptors (R7p %), the number of replicate slides counted, the standard deviation among slides and the mean adjusted for Wolbachia infection and inversions. Only these per-line values were published (no per-slide counts), so the study has a summary dataset only. The lines are given as bare numbers (e.g. 21 = DGRP_021); all of them are DGRP lines of DGRPool. No journal version of the preprint was found (November 2026).
Phenotype: each R7 photoreceptor expresses one of two UV opsins, Rh3 ("pale" R7, R7p) or Rh4 ("yellow" R7, R7y), a stochastic cell-fate choice controlled by
spineless (about 30% pale R7s in the
cn bw laboratory strain). Adult eyes of the homozygous DGRP lines were dissected from isolated heads (about 20 eyes per sample), dissociated into ommatidia on a slide and stained with antibodies against Rh3 and Rh4, and the percentage of pale R7 cells (R7 cells expressing only Rh3; cells expressing Rh4 were scored as yellow) was counted by direct visualization under a fluorescence microscope, on 3 to 8 replicate slides per line (flies raised at 25 °C on cornmeal food). The paper does not give the sex (or age) of the flies, so the sex is NA.
The summary dataset has:
-
mn_R7pale and
sd_R7pale (%): mean and standard deviation over replicate slides of the percentage of pale R7s (the means are rounded to 0.01 in Table S1). They range from 25.01% (DGRP_801) to 61.43% (DGRP_374), with a mean of 41.1%, as in the paper ("between 25% and 61% with a mean of 41%").
-
mn_R7paleAdj (%): the line mean adjusted by the DGRP2 webserver for Wolbachia infection and the five major inversions (model intercept plus the line residual), the phenotype of the authors' GWAS. It is reproduced from the unadjusted means, the DGRP Wolbachia and inversion genotypes and the effects listed in Supplemental Table S9 to within 0.005 percentage points (the rounding of the published means). In(3R)Mo has the largest effect (+18.5 points for the three homozygous lines DGRP_374, DGRP_437 and DGRP_820; P = 0.00085), which the authors discuss (Figure S3). DGRP_377 and DGRP_786 have the same adjusted value (40.948%, the model intercept) because each is the only tested line with its inversion genotype (In(2R)NS heterozygous and In(3R)P homozygous, respectively), so its residual is zero: this is not a copy error.
-
NSlides_R7pale (obsolete, design information): the number of replicate slides counted per line.
The broad-sense heritability (R
2 of a one-way ANOVA of the slides on the line) computed from Table S1 is 29.7% for the adjusted values, as in the paper (29%), but 43.5% for the unadjusted values, whereas the paper gives 49%. This cannot be checked further without the per-slide data.
GWAS: DGRP2 webserver (FaST-LMM on variants with MAF ≥ 5% in the 46 lines; P < 10
-5): 42 variants near 28 candidate genes (Supplemental Table S2). In an RNAi screen of the candidates, 12 of the 37 RNAi lines tested reduced the proportion of pale R7s.
Not uploaded: the
cn bw control strain (not a DGRP line); the RNAi validation screen (Table S5:
GMR-GAL4 crossed to UAS-RNAi lines and their controls, not DGRP lines); the RNA-seq of 40 h APF pupal retinas of
sevenless mutants and
w1118 controls (Tables S6, S7, S10; GEO GSE220550, not DGRP lines); the candidate-gene and network tables (S3, S4); and Table S8, a re-analysis with the DGRP2 webserver of the % SsON (yellow) R7 data of Anderson et al. 2017, which is
DGRPool study 180 (F1 males of 203 DGRP lines crossed to a
spineless deficiency). Over the 46 lines shared with study 180, mn_R7pale correlates negatively with its % SsON R7s (Pearson r = −0.39, P = 0.007), as expected since pale R7s are the SsOFF cells.