Bioinformatic analyses of mammalian 5'-UTR sequence properties of mRNAs predicts alternative translation initiation sites

Table 5 5'-UTR Sequence Properties Utilized for Analyses by the Artificial Neural Network (ANN)

Parameters	Description	Range
Consensus Sequence Patterns
Pattern #1 (C/(G/C)	Position -6/-7	Yes/No
Pattern #2 (Kozak)	Position -3/+4	Yes/No
Consensus Sequence Patterns
5'-UTR Length	Length of 5'-UTR	80 to 2000 bp
ORF Length	Length of ORF from the annotated start codon to the stop codon	350 to 5000 bp
Start Codon	Frequency of aTIS in training set	0 to1
Number of AUGs	Number of upstream AUGs from aTIS	0 to 19
G/C Ratio	Normalized ratio of G to C	-1 to 1
Secondary Structure
Free Energy	50 bp UnaFold	-40 to 0
Secondary Structure	UnaFold	0 = stem, 3 = loop

Properties of 5'-UTR sequences of mRNAs using aTIS are shown according to their application in the ANN. The derivation of each feature is shown, as well as the range of representation to the ANN for training and testing. These features are implemented in the ANN analyses which includes refined representations of secondary structure.

ISSN: 1471-2105