Antisense DNA parameters derived from next-nearest-neighbor analysis of experimental data

Area of Science:

• Molecular Biology
• Genetics
• Bioinformatics

Background:

• Antisense DNA technology offers gene expression control by targeting specific gene sequences.
• Predicting effective antisense DNA targets traditionally relied on identifying sequence motifs correlated with gene inhibition.
• Nearest-neighbor (NN) thermodynamic models, derived from in vitro experiments, do not fully capture in vivo antisense DNA effects due to overlapping nucleotide interactions.

Purpose of the Study:

• To develop and validate a next-nearest-neighbor (NNN) analysis model for predicting in vivo antisense DNA efficacy.
• To account for overlapping nucleotide sequence effects in antisense DNA inhibition.
• To improve the accuracy of identifying effective antisense DNA targets for gene expression regulation.

Main Methods:

• Utilized singular value decomposition (SVD) to analyze experimental data from phosphorothioate-modified antisense DNAs (S-DNAs).
• Applied a NNN model to fit in vivo data from 112 experiments involving four gene targets and two cell lines.
• Derived NNN inhibition parameters representing all possible overlapping triplet interactions within cellular targets.

Main Results:

• The NNN model adequately fitted the experimental data, providing parameters for 49 distinct overlapping triplet interactions.
• NNN triplet parameters demonstrated superior performance compared to NN doublet parameters in modeling in vivo antisense DNA effects.
• The model successfully incorporated parameters for specific mRNA targets and cell lines, showing flexibility.

Conclusions:

• The NNN analysis methodology effectively derives in vivo antisense DNA inhibitory information by considering overlapping nucleotide effects.
• This approach provides a more robust method for calculating antisense inhibitory parameters for any mRNA sequence compared to motif tallying.
• The derived NNN parameters' applicability is currently limited by the size of the experimental database used for their derivation.