Determining the correct structure of a protein based on its sequence remains a difficult task, and our protein engineering team is working towards this goal. Most structure prediction methods generate a large number of possible candidates, and the ultimate challenge is to select the best among them. With years of experience, we provide customized protein model ranking and analysis service to precisely meet customer requirements.

Introduction of Protein Model Ranking

Over the last two decades, various protein 3D structure prediction methods have been developed and much progress has been made in this field. Typically, a large number of predicted decoy models are generated for a given protein sequence, and correctly ranking these models and selecting the best predictive model from a pool of candidates remains a challenging task. Currently, researchers have developed many methods to rank the quality of model structures, such as probability estimation, the MQAPRank program, DeepQA, the QMEAN server, etc. In addition, scoring functions have been developed that aim to estimate the expected accuracy of the model. The first class of scoring functions relies on the analysis of individual models based on incoming or physical chemistry criteria. The second class uses full-to-all structural comparisons of models to derive quality scores from the information contained in the set of models for a given sequence.

Fig 1. The Deep Belief Network architecture for DeepQA. ( (Cao R, et al., 2016)

Services

The selected model from the generated model pool is the so-called model ranking problem. Model quality estimation is an important part of protein structure prediction, as the accuracy of a model ultimately determines its usefulness for a particular application. As a leading service provider of protein engineering, Creative BioMart has successfully developed an advanced technology platform for sequencing and analysis of protein models. Based on this platform, we can evaluate a single model or multiple models, each of which is assigned a quality score that is used to rank the structure. In addition, per-residue error estimates are provided and visualized in multiple ways, allowing customers to examine protein models in greater detail.

Our scientists are committed to developing strategies to select good models from the pool of generated models to help you with model ranking and analysis. These strategies have their merits, and we will select them based on the situation and the availability of information.

• Single approach: only using the input model to evaluate model quality. And three conceptual approaches are often used: physical models, statistical models, and comparisons between predicted attributes and attributes extracted from decoy models.
• Quasi-single approach: taking some high-quality models as a reference and evaluate subsequent models by comparing them with the reference model.
• Clustering (or consensus) approach: clustering algorithms are typically used to calculate the average structural similarity score of a model to other models as its model quality.

Creative BioMart is committed to providing high-quality protein model ranking and analysis services to global customers, 3D models of proteins are compared with the quality of the models obtained through the scoring system. We will work with you to develop the most appropriate strategy and provide the most meaningful data for your research for accelerating the research of life sciences. If you are interested in our services, please do not hesitate to contact us for more information.

References

1. Cao R, Bhattacharya D, Hou J, et al.. (2016) DeepQA: improving the estimation of single protein model quality with deep belief networks[J]. BMC bioinformatics. 17(1): 1-9.
2. Jing X, Dong Q. (2017) MQAPRank: improved global protein model quality assessment by learning-to-rank. BMC Bioinformatics. 18(1): 275.

• Detection of Errors in Protein Models
• Folding and Molecular Simulation Service
• Molecular Docking Service
• Protein Conformational Isomerism Detection and Analysis Service
• Protein Crystallization Service