Sites identified as signatures of b12 sensitivity using any of the three signature-defining approaches: contingency table, CMI, and ensemble machine learning method.

1The CMI approach does not provide specific information regarding which amino acids give rise to the signal, although particularly distinctive substitutions can be seen by examining the data (Sup. Fig. 1). A “Yes” in the CMI column means the site was associated with b12 sensitivity or resistance.
2The Fisher's exact contingency table is based on specific amino acids or sets of amino acids, such that the amino acids associated with signature sites are explicit; and amino acids associated with b12 resistance are underlined, whereas amino acids associated with b12 susceptibility are not underlined.
3The recurrent top splits in the decision trees (the number in parentheses indicates how many times it was found) provide information about the key signature amino acid substitutions. The exclamation point (!) means “not” in these tables and figures, thus E→!E means that “E” is found in the immediate ancestral state of the sequence, and is “not E” in the sequence.