Average properties of predicted APRs in different datasets.

*Average length of sequences that contain at least one TANGO/WALTZ predicted APRs.
†Average proportion of APR residues in the sequences contained in different datasets was computed as described in methods.
Sequences that do not contain APRs were excluded from both the average length and average proportion calculations. Standard deviations (σ) are reported for all averages. Standard error (SE) of the mean can be computed as σ/√(number of sequences; see Table 2) with 95% confidence intervals (average ± SE*1.96).