Disulfide-rich peptides are wide spread in nature and as hormones and toxins play key biological roles for many organisms. Some families of disulfide-rich peptides are so large and diverse that they can be considered natural combinatorial libraries for drug discovery. Cystine residues result from the formation of disulfide bonds between pairs of cysteine residues. This cross linking of the protein backbone is essential for the structure and activity of peptides and proteins. The conformation of a cystine side chain can be described using five dihedral angles, χ1, χ2, χ3, χ2’, and χ1’, with cystines favouring certain combinations of these angles. 2D NMR spectroscopy is ideally suited for structure determination of disulfide-rich peptides because of their small size and constrained nature. Notably NMR spectroscopy cannot be used to directly determine the disulfide connectivity, and only limited information can be derived for the cystine side chain conformation. This often limits the precision and accuracy leading to ambiguity in the deduced 3D structures. Disulfide-rich peptides have proven to be promising drug candidates in a number of fields, either as bioactive leads or scaffolds. Resolving accurate structures can accelerate the drug development process through the introduction of a rational design component. Building a peptide atlas will expand the understanding of SARs and allow for the design of novel analogues with targeted receptor interactions.This thesis expands the understanding of the association between cystine conformations and the surrounding hemi-Cys structure through an extensive analysis of >19,000 cystines in reported structures. Using a classification system for conformations based on all five dihedral angles, a complex relationship with a number of structural features was discovered. Of particular importance were the types of secondary structure linked by cystines. This not only influenced the overall conformation, but individual χ angles were found to adopt specific sub-configurations.Using a database of NMR chemical shifts combined with crystallographic structures, a method called DISH that uses support vector machines (SVM) was developed to predict the dihedral angles of Cys side chains. It is able to successfully predict χ2 angles with 91% accuracy, and has improved performance over existing prediction methods for χ1 angles, with 87% accuracy. For 81% of Cys residues, DISH successfully predicts both the χ1 and χ2 angles. By revisiting published solution structures of peptides determined using NMR spectroscopy, impact of additional cystine dihedral restraints on the quality of 3D models was assessed. DISH improves the resolution, highlighting the potential for improving the understanding of SARs and rational development of peptide drugs.Most computational methods that predict Cys connectivity are based on amino acid sequence homology and have therefore limited performance for novel peptides. Given that cystines do favour specific conformations and that the dihedral angles were able to be predicted based on chemical shifts, these concepts were extrapolated to implement a computational approach called CrossLink that predicts cystine connectivity from chemical shift inputs. It uses a SVM to first predict the χ3 dihedral angle of Cys residues using NMR chemical shift data and without knowledge of the connectivity. This angle as well as experimental NMR chemical shifts is then fed to a deep neural network that suggests the disulfide connectivities. CrossLink has a global accuracy of 82% for predicting the connectivity of Cys pairs and an accuracy of at least 90% for predictions that are suggested to be confident.The experimental applications of CrossLink and DISH on a series of novel conotoxins were used to demonstrate their impact. Conotoxins are disulfide-rich peptides found in the venom of the Conus genus. Due to their exquisite potency and high selectivity for a wide range of voltage and ligand gated ion channels they are extremely attractive drug leads in neuropharmacology. Of particular note has been their use for the treatment of neuropathic pain as antagonists of the CaV2.2 channel. The new ω-members CVIE and CVIF_R10K are promising drug candidates due to an increased therapeutic index. Furthermore, it was recently found that cone snails have the capability to rapidly switch between venom types with different proteome profiles based on predatory or defensive stimulation. A novel conotoxin belonging to the I3-superfamily, G117, was identified as the major component of the predatory venom of Conus geographus. Of significant biological interest, its activity is currently unknown. To gain insights into activity and develop SARs the 3D structures of CVIE, CVIF_R10K and G117 were resolved by 2D NMR spectroscopy.Two unique structural features were identified in CVIE and CVIF_R10K that were hypothesised to result in the increase of the therapeutic index. The first was the Ser12 residue, replacing a key hydrophobic interaction with the CaV2.2 channel present in other ω-conotoxins. The second feature was the presence of the larger hydrophobic Leu22 residue in CVIF_R10K. It was conjectured that this results in steric hindrance of surrounding basic side chains known to be essential to activity. G117 is the first structure to be reported for the I3-superfamily. The 32 amino acid peptide is comprised of eight Cys residues with a Cys XI framework. Using CrossLink in conjunction with experimental data, the disulfide connectivity was identified as forming an ICK+1 motif. Despite little conservation between the sequences, G117 had a high degree of structural conservation and comparable side chain positions to the ι-RXIA conotoxin, an agonist for NaV channels.Overall this thesis reports methods to increase the accuracy and precision of disulfide-rich peptide structures that are resolved by NMR spectroscopy. Expanding the size and quality of the structural atlas of peptide toxins will progress the understanding of the relationships between structures and activity. This is essential as the drug development pipeline transitions towards a rational design process to satisfy increasing regulatory and social demands.