Kindly ask how the EVEscape score from fig4d heatmap and the evescape.org was calculated?

[ipplol]

I followed the scripts to get the full_spike_evescape.csv.
The "evescape", "evescape_pos", and "evescape_sigmoid" do fit with the result of score_pandemic_strain.py as I downloaded from https://marks.hms.harvard.edu/evescape/strain_scores_20230318.zip

But they seem to be much smaller than the score, especially the in-silico DMS result, on the evescape.org website.
Kindly ask if there are any additional steps of transformation or normalization？