Table 1.

Results on estimating translation efficiency (TE) from the TITER prediction scores, using a linear regression model with different combinations of features

Feature set	Feature coefficients in regression (mean±SD)	Spearman’s correlation r	P value	MSE
aTIS	1.833±0.034	0.234	1.077e−72	3.081
aTIS+uTIS	1.774±0.040, –0.126±0.005	0.245	3.801e−79	3.063
aTIS+AUG	1.709±0.038, –0.029±0.005	0.234	2.712e−72	3.078
aTIS+uTIS+lofUTR	1.744±0.037, –0.122±0.005, –0.0001±3.634e–5	0.236	2.228e−73	3.063

‘MSE’ denotes the mean square error. ‘aTIS’ represents the predicted TISScore for the aTIS, ‘uTIS’ represents the sum of TISScore values of all the eligible uTISs, ‘AUG’ represents the number of AUGs in the upstream region of the aTIS, and ‘lofUTR’ represents the length of the 5’ UTR. The mean and the standard deviation (SD) of the feature coefficients in the regression model in a 10-fold cross-validation procedure were calculated.