Table 1.

Head-to-head model comparisons

(a) Comparing M2 with other EMASE models
	M1	M3	M4
% of Genes:
M2 is better	11.5 ± 0.2	1.1 ± 0.1	2.3 ± 0.2
The other is better	3.3 ± 0.1	0.6 ± 0.1	2.0 ± 0.1
% of Isoforms:
M2 is better	16.9 ± 0.2	18.2 ± 0.2	18.7 ± 0.2
The other is better	11.0 ± 0.2	12.2 ± 0.2	13.1 ± 0.2
(b) Comparing M2 with other existing methods
	RSEM	kallisto	Unique	WASP
% of Genes:
M2 is better	3.7 ± 0.2	7.1 ± 0.3	6.2 ± 0.1	26.2 ± 0.8
The other is better	2.9 ± 0.1	4.4 ± 0.2	2.6 ± 0.1	5.1 ± 0.2
% of Isoforms:
M2 is better	19.8 ± 0.3	22.2 ± 0.2	13.6 ± 0.2	N/A
The other is better	13.2 ± 0.2	14.0 ± 0.2	5.6 ± 0.1	N/A

Note: For each pair of methods, we consider only those genes for which one or both estimates of ASE deviate by 5% or more from simulated truth. We then tabulated the proportion of genes or isoforms for which one model is closer to the truth. The pairwise comparisons are summarized relative to EMASE model M₂. WASP does not provide isoform-level estimation. See Supplementary Figures S2–S6.