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

Single-stranded DNAs (ssDNAs) play major biological functions and represent an interesting biotechnological tool. They constitute a compelling alternative to RNA, because of their greater stability as compared to the latter. As for RNA, ssDNAs function depends on the specific foldings they adopt. Therefore, information about ssDNAs 3-dimensional (3D) structures is fundamental to investigate their functions. In this context, in silico 3D structure prediction can facilitate ssDNA design. This task can be addressed indirectly, by using the tools for RNA structure prediction and then converting the output in the ssDNA format, or one of the few tools capable of directly handling ssDNA. This study assessed 3 indirect RNA 3D structure prediction methods (RNAComposer, SimRNA, and Vfold3D), based on their performance in the Critical Assessment of Structure Prediction and one direct DNA prediction tool (3dDNA) to evaluate their performances in modeling ssDNAs. At this scope, a dataset of 97 experimentally determined ssDNA structures, including challenging motifs such as G-quadruplex, was built. Various metrics, namely, Root Mean Square Deviation, Global Distance Test Total Score, and Interaction Network Fidelity, were employed to benchmark the accuracy of the predictions. The 3 indirect tools showed similar and moderate performances, while the direct tool provided better results. Nevertheless, they all performed poorly in modeling G-quadruplexes and structures containing motifs increasing the intrinsic flexibility of ssDNA. Despite the recent efforts in the prediction of the 3D folding of ssDNAs, improvements in method are still needed. This should involve taking into account the conformational variability of this kind of molecules and paying attention to their specific 3D motifs.