Average aggregation propensities of sequences in different datasets.

*TANGO/WALTZ aggregation propensity of a sequence was computed as its total TANGO/WALTZ aggregation score normalized by the number of residues in the sequence.†Dataset SF49500 was obtained by randomly scrambling each sequence in F495 a hundred times (see Methods for details).‡Datasets F1 and F2 were obtained by dividing F495 into two sets such that their amino acid compositions are significantly different (see Methods for details). Standard deviations (σ) are reported for all averages.Standard error (SE) of the mean can be computed as σ/√(number of sequences) with 95% confidence intervals (average aggregation propensity ± SE*1.96).