As already mentioned, it doesn't seem likely the SNPs found in this study would have a good diagnostic capability, given how the original gene differential expression study from 2008 that it's based on didn't validate. (That should have been no surprise given the small patient sample size and their heterogeneity.) But I thought I'd look at it for completeness anyway.
Of the 10 of 21 SNPs deemed significant in Table 2, 23andMe appears to give only 3 of these (at least on the v3 chip platform that I was tested on a couple years ago).
Worse, some of the data for those 3 SNPs looks problematic (see the Notes below), so this exercise was probably not very useful:
Gene | rs# | (risk allele) | my results
FAM126B rs11895568 NA
TCF3 rs1860661 (A) GG -/- (Note 1)
EIF3A rs10787901 NA
UBTF rs2071167 ("A") CC -/- (Note 2)
METTL3 rs3752411 (A) AG +/- (Note 3)
SORL1 rs3737529 NA
BMP2K rs3775516 NA
BMP2K rs1426139 NA
IL6ST rs1373998 NA
PEX16 rs3802758 NA
Note 1: Their allele frequency data for this SNP is way out of line (much too high) compared to other sources, so I'm guessing they've made some kind of scribing error here - either the rs# is incorrect or they're giving data belonging to another rs#. (They also give the major allele as the risk allele, which wouldn't inherently be a problem except for the seemingly bad frequency data.)
Note 2: They apparently use minus strand labeling here (while using positive labeling for the other two SNPs that 23andMe gives data for) - or more likely they've made another scribing error (they may just need to replace "AG" with "CT") . The allele frequencies seem within reason.
Note 3: Their allele frequency data again looks out of line (too low) for this SNP.
We might try to inquire with the authors about the seeming discrepancies, or perhaps an Erratum will be published to correct it.
(I didn't look at the 148 SNP alleles that this same study also found are supposed to associate with different CFS/ME subtypes. For those rs#'s, see this
Supplementary Table.)