TukeyHSD does not match the box plot

Question

[Ana]

Hi all,

I have done anova followed by Tukey for multiple comparisons to compare the depth of coverage between different genotypes (homozygote for reference allele, homozygote for alternative allele and heterozygote) on different chromosomes. The strange thing is that, Tukey shows sifinificat difference between different genotype types but this dos not match the box-plots that I have produced

                        diff          lwr         upr        p adj chrom hom_ref_count hom_alt_count het_count
hom_alt-het      -0.27288393  -0.44718368 -0.09858418 7.131898e-04    10          1360          5827      4437
hom_ref-het       0.23482928  -0.03631045  0.50596900 1.050828e-01    10          1360          5827      4437
hom_ref-hom_alt   0.50771321   0.24426968  0.77115673 1.883648e-05    10          1360          5827      4437
hom_alt-het4     -1.22787653  -1.26328218 -1.19247089 0.000000e+00     6        224270        115659    178236
hom_ref-het4     -0.29767089  -0.32742635 -0.26791543 0.000000e+00     6        224270        115659    178236
hom_ref-hom_alt4  0.93020564   0.89626011  0.96415118 0.000000e+00     6        224270        115659    178236
hom_alt-het5      0.22126344  -0.34114745  0.78367432 6.258311e-01    12           524           391       607
hom_ref-het5     -0.40844411  -0.92562501  0.10873680 1.530304e-01    12           524           391       607
hom_ref-hom_alt5 -0.62970754  -1.20930663 -0.05010845 2.934662e-02    12           524           391       607
hom_ref-het7     -1.46804924  -2.40773839 -0.52836008 2.239110e-03    15           404             0       370

You can find box-plots here, enter link description here

as you see there seems to be no significant differences

I wonder what is wrong or what should I do?any idea?

[Alex Reynolds]

Perhaps make notched box plots to more directly do comparisons of medians between two groups. https://www.jstor.org/stable/2683468?seq=1#page_scan_tab_contents