Amino acid distributions at individual positions are not correlated with HAD.
A. Amino acid frequencies in the brain dataset plotted as distributions totaling 100% for each class (HAD, non-HAD). The weights of individual sequences are normalized by patient sequencing depth. B. Percentage of sequences of each class (HAD, non-HAD) matching the amino acid requirements of signature 1_04 at each position individually, and for the complete signature. Bars represent only matching sequences and thus do not sum to 100%.