Efficient context-dependent model building based on clustering posterior distributions for non-coding sequences

Table 3 Split calculation for the melting model-switch integration.

δβ	Integration interval	Q.E.	σ_d	σ_s	total error
0.0001	1.00–0.99	22.0	1.2	0.6	2.2
0.0001	0.99–0.98	10.2	0.0	0.1	0.2
0.0002	0.98–0.96	19.2	0.0	0.1	0.2
0.0002	0.96–0.94	18.5	0.0	0.1	0.1
0.0002	0.94–0.92	18.1	0.0	0.0	0.1
0.0002	0.92–0.90	17.8	0.0	0.0	0.1
0.001	0.90–0.80	87.5	0.0	0.2	0.3
0.001	0.80–0.70	84.5	0.0	0.1	0.2
0.001	0.70–0.60	82.7	0.0	0.1	0.2
0.001	0.60–0.50	80.8	0.0	0.1	0.2
0.001	0.50–0.40	77.9	0.0	0.2	0.3
0.001	0.40–0.30	74.5	0.0	0.2	0.3
0.001	0.30–0.20	69.1	0.0	0.2	0.4
0.001	0.20–0.10	58.1	0.1	0.4	0.7
0.0002	0.10–0.08	8.0	0.0	0.1	0.2
0.0002	0.08–0.06	5.6	0.0	0.1	0.2
0.0002	0.06–0.04	1.3	0.0	0.2	0.3
0.0002	0.04–0.02	-8.7	0.1	0.2	0.5
0.0001	0.02–0.01	-15.5	0.1	0.2	0.4
0.0001	0.01–0.00	-57.8	0.6	0.4	1.2
Total	0.00–1.00	653.7	2.3	3.6	8.2
Composite Run log Bayes Factor: 653.7
Composite Run Confidence Interval: [645.5; 661.9]

Using a constant number of Q = 200 iterations per β, the contribution of each integration interval to the Bayes Factor value was calculated on a separate processor. This leads to an improved approximation of the contribution for the intervals [1.0; 0.9] and [0.1; 0.0] and also decreases the width of the confidence interval from 43.9 to 16.3. The decrement δβ was allowed to change and Q = 200 iterations were performed for each value of β. Q.E. is the quasistatic estimator for each integration interval with discrete and sampling error denotes by σ_dand σ_s, respectively.

ISSN: 2730-7182