Table 2b. Concordance of splice-prediction programs to information theory-based tools for splicing regulatory proteins.

	ESEfinder 3, 4	Rescue-ESE	Ex Skip 3, 4	ESEsearch	PESX
ESEs (all types)	9/15	3/4	4/14	2/3	1/1
Neut. Mut.	4/4	1/1	3/3	-	-

Concordance was assessed from the analysis of variants from 36 publications which used IT-based tools and a secondary predictive method. Each value corresponds to the number of variants that were concordant with IT-based tools versus the total number of variants for each category. ¹ – includes Vreeswijk et al. (2008), which may not have properly reported predicted cryptic sites, as they did not report any cryptic sites predicted by ASSA beyond the default window size (54 nt) from the mutation. ² – predictions made using the SSF-like algorithm in the Alamut splicing prediction module were combined with the SSF category (SSF is no longer supported). ³ – Aissat et al. (2013) contributes highly to the discordance of these programs, and may be due to improper reporting/analysis. ⁴ – Mutations predicted by alternate program to affect SR protein to which ASSEDA has no model (i.e. 9G8) were not included in statistics.

MES – MaxEntScan; BDGP – Splice Site Prediction by Neural Network, NNSplice; NG2 – NetGene2; HSF – Human Splice Finder; SSF – Splice Site Finder; SSqF – Splicing Sequences Finder; GS - GeneSplicer; SV – SpliceView; SP – Splice Predictor; SS - Shapiro-Senapathy; GenS – GenScan; ASD - ASD-Intron analysis; GeneS – GeneScan; GM – GeneMark; PESX - Putative Exonic Splicing Enhancers/Silencers.