Statannotator indexing issue

[jyaacoub]

For some figures, it seems that the tool isnt working properly:

MSE loss for kiba edge types:

Figure

Issues:

• simple-anm is not significant and that bar should belong to binary-anm
• binary-anm bar is actually ** and this bar should be fore binary-af2

Actual Values:

MSE:
    binary vs. simple:  Mann-Whitney-Wilcoxon test two-sided, P_val:3.095e-01 U_stat=7.000e+00
   *anm vs. binary:     Mann-Whitney-Wilcoxon test two-sided, P_val:7.937e-03 U_stat=2.500e+01
    af2 vs. anm:    Mann-Whitney-Wilcoxon test two-sided, P_val:2.222e-01 U_stat=6.000e+00
   *af2 vs. binary:     Mann-Whitney-Wilcoxon test two-sided, P_val:3.175e-02 U_stat=2.300e+01
    anm vs. simple:     Mann-Whitney-Wilcoxon test two-sided, P_val:1.508e-01 U_stat=2.000e+01
    af2 vs. simple:     Mann-Whitney-Wilcoxon test two-sided, P_val:3.095e-01 U_stat=1.800e+01

MSE loss for PDBbind feature types:

Figure

Issues:

• nomsa-msa bar belongs to nomsa-shannon
• msa-esm bar belongs to shannon-esm

Actual Values:

MSE:
    msa vs. ESM:    Mann-Whitney-Wilcoxon test two-sided, P_val:6.905e-01 U_stat=1.500e+01
    nomsa vs. msa:  Mann-Whitney-Wilcoxon test two-sided, P_val:5.556e-02 U_stat=3.000e+00
    nomsa vs. ESM:  Mann-Whitney-Wilcoxon test two-sided, P_val:9.524e-02 U_stat=4.000e+00

   *shannon vs. ESM:    Mann-Whitney-Wilcoxon test two-sided, P_val:1.587e-02 U_stat=2.400e+01
   *nomsa vs. shannon:  Mann-Whitney-Wilcoxon test two-sided, P_val:7.937e-03 U_stat=0.000e+00
   *shannon vs. msa:    Mann-Whitney-Wilcoxon test two-sided, P_val:3.175e-02 U_stat=2.300e+01