Comments from curator
Phenotyping data comes from the authors' figshare item
Susceptibility of different Drosophila melanogaster DGRP lines to three sigma viruses (
doi:10.6084/m9.figshare.6743354, cited in the
paper's Data availability statement), file
DGRP_data_out.csv: 95 viral loads of 32 DGRP lines, with the
p62/
CHKov1 genotype class of each line,
but no column saying which virus each value belongs to (see below).
Experiment: 32 DGRP lines chosen by their alleles of the two known resistance genes to the sigma virus of
D. melanogaster,
p62 (
ref(2)P) and
CHKov1 (11 lines susceptible for both, 13 resistant for
CHKov1 only, 8 resistant for both; no line is resistant for
p62 only), were injected with three sigma viruses (rhabdoviruses): DMelSV, the natural (coevolved) virus of
D. melanogaster, and DAffSV and DObsSV, isolated from
D. affinis and
D. obscura. The viral RNA load was measured 15 days post infection by quantitative RT-PCR relative to the fly reference gene
RpL32, as log
2 viral load (the negative of ΔCt, the difference between the cycle thresholds of the virus and
RpL32, after correcting the Ct values for qRT-PCR plate). 1,869 flies were injected, a mean of 18 per line (range 3-22). The sex of the DGRP flies is not stated (the other
D. melanogaster experiments of the paper used males), so the sex is NA.
Raw dataset 1 (32 lines): each value is one sample, the flies of one line injected with one virus and measured together; there is one sample per line and virus. The three viruses are side by side (
ViralLoad_DMelSV,
ViralLoad_DAffSV,
ViralLoad_DObsSV, log
2, no unit): values on the same row come from different flies. Values are copied as in the authors' file.
How the virus of each value was recovered: the file only has the columns line, genotype class and viral_load, and the rows of a line are not in a fixed virus order.
Figure 3 plots the same 95 values per genotype class, coloured by virus (red DMelSV, black DAffSV, grey DObsSV). Each value of the file was matched to the colour of the point drawn at its height in its genotype class (no value was taken from the figure). 18 lines matched only one way; the other 14 were resolved jointly, as the assignment of all their values that best reproduces the points of the figure. The result reproduces the statistics printed in the paper (effect of genotype class on viral load: DMelSV F
2,28 = 13.2, DAffSV F
2,29 = 4.9, DObsSV F
2,29 = 5.7) and the nine medians drawn in Figure 3. Only 6 of the 36,864 possible assignments of the 14 ambiguous lines reproduce all of these; they differ from the one used here only by swapping the DAffSV and DObsSV values of DGRP_049, DGRP_367, DGRP_386 or DGRP_595, which are within 0.15 log
2 of each other.
Inconsistency in the source: the file has 4 rows for line 158 (two of them DMelSV: -10.06 and -8.05) and only 2 rows for line 105 (no DMelSV), both in the class susceptible for both genes. One of the two DMelSV values of line 158 is probably line 105's, but the file does not say which, so the values are kept as published: DGRP_158 has two DMelSV samples (DGRPool mean -9.06) and DGRP_105 has no DMelSV value. DGRP_861 has no DMelSV value either (the paper's DMelSV test has 31 values).
Not uploaded: the genotype class of each line (column ref2p_chkov: 0_0 susceptible for both genes, 0_1 resistant for
CHKov1 only, 1_1 resistant for both), which is a DGRP genotype and not a measured phenotype. The main experiments of the paper do not use the DGRP: full-sib families from wild populations of four
Drosophila species (
figshare 6743339) and QTL mapping in 377 DSPR lines (
figshare 7195751). They show more genetic variation in susceptibility to the coevolved virus, largely due to a few major-effect variants; in the DGRP lines, the resistant alleles of
p62 and
CHKov1 reduced the load of DMelSV much more than that of DAffSV and DObsSV.