Predicting Single-Stranded DNA Oligonucleotides 3D Structures: An Open Issue

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology
• Biotechnology

Background:

• Single-stranded DNAs (ssDNAs) are vital for biological functions and biotechnological applications due to their stability and functional foldings.
• Understanding the 3D structures of ssDNAs is fundamental for investigating their roles and for designing novel ssDNA molecules.
• In silico 3D structure prediction offers a powerful approach to facilitate ssDNA design and analysis.

Purpose of the Study:

• To evaluate the performance of existing 3D structure prediction tools for single-stranded DNAs (ssDNAs).
• To compare the accuracy of indirect methods (using RNA prediction tools) versus direct methods (using DNA-specific tools) for ssDNA modeling.
• To identify limitations in current prediction methods, particularly for complex ssDNA structures.

Main Methods:

• A dataset of 97 experimentally determined ssDNA structures was compiled, including challenging motifs like G-quadruplexes.
• Three indirect RNA 3D structure prediction tools (RNAComposer, SimRNA, Vfold3D) were assessed.
• One direct DNA prediction tool (3dDNA) was evaluated alongside the indirect methods.
• Performance was benchmarked using metrics such as Root Mean Square Deviation (RMSD), Global Distance Test Total Score (GDT_TS), and Interaction Network Fidelity (INF).

Main Results:

• Indirect RNA prediction tools demonstrated moderate and comparable performance in modeling ssDNAs.
• The direct ssDNA prediction tool (3dDNA) generally provided superior results compared to indirect methods.
• All evaluated tools exhibited poor performance in accurately modeling G-quadruplexes and ssDNA structures with high intrinsic flexibility.
• Current prediction methods require improvement to account for conformational variability and specific 3D motifs in ssDNAs.

Conclusions:

• While direct prediction tools show promise for ssDNA modeling, current methods are insufficient for accurately predicting complex structures.
• Further advancements are necessary to enhance the prediction of ssDNA 3D structures, especially for motifs like G-quadruplexes.
• Future research should focus on incorporating conformational variability and specific structural motifs into ssDNA 3D structure prediction algorithms.