Finding sequence motifs with Bayesian models incorporating positional information: an application to transcription factor binding sites

Table 2 The correlation coefficients for the TSS Tompa data subsets

Data Subset	Without positional information	With positional information	Improvement
hm01r	-0.012	-0.007	0.005
hm02r	-0.009	-0.007	0.002
hm03r	-0.037	0.386	0.423
hm04r	-0.008	-0.005	0.003
hm05r	-0.031	-0.019	0.012
hm06r	-0.014	0.156	0.170
hm07r	-0.015	-0.015	-0.001
hm08r	-0.012	0.574	0.586
hm09r	-0.011	0.358	0.369
hm10r	-0.019	0.083	0.102
hm11r	-0.028	-0.012	0.016
hm13r	-0.015	-0.016	-0.001
hm14r	0.204	-0.018	-0.222
hm15r	-0.011	-0.012	-0.002
hm16r	-0.011	-0.006	0.005
hm17r	-0.015	-0.012	0.004
hm18r	-0.018	0.094	0.112
hm19r	-0.010	-0.007	0.003
hm20r	-0.026	0.046	0.073
hm21r	0.401	0.384	-0.016
hm22r	-0.020	-0.020	0.000
hm24r	-0.016	-0.010	0.006
hm26r	-0.016	0.099	0.115
Combined CC	-0.008	0.101	0.109

Table 2 shows the correlation coefficients for A-GLAM's predictions on the 23 subsets of the TSS Tompa dataset. The column, "Improvement", quantifies the effect of positional information on predictions, by showing the difference between the correlation coefficients in the second and third columns, "Without Positional Information" and "With Positional Information".

ISSN: 1471-2105