Researches On Transmembrane Protein Fold Recognition

Transmembrane proteins (TMP) are special proteins locating on the biomembranes. Differing with the soluble proteins which exist in the soluble environments inside the cells, they cross the biomembranes through particular secondary segments and locate on them stably spending their whole life-cycle. That is the reason why they are named transmembrane protein. They are highly important in various biological processes and pharmaceutical developments, such as transporters and receptors which respond the transferring of materials and signals between the two sides of biomembranes. Without the participant of transmembrane proteins, many microscopic life processes cannot be realized and the pathway will be broken. Due to their indispensable functional contributions, they are also the biological targets for most drugs currently on market, and many potential drugs will be designed and produced relied on transmembrane proteins in future. Therefore, studying on transmembrane proteins is vital important for exploring the mechanism of life, comprehending the rule of life processes, solving the problem of complex human diseases and pushing drug designations.It well known that the functions of proteins are decided by their spatial structures, so that structural research of protein becomes a major approach to discover their functions. With the rapidly developing biological technologies, high-resolution spatial structures of soluble proteins are increasing in amount through experimental methods, which thereby provide reliable data for deep understanding their functions and mechanisms, and further support the protein structure prediction and function prediction. However, the spatial structures of transmembrane proteins are hard to obtain using such approach as soluble proteins due to their special biochemical and biophysical properties. Meanwhile, huge amount of protein sequence have been released by sequencing technologies recently. Hence, spatial structures of transmembrane proteins are expected to be predicted from their amino acid sequences. But, the methods that predict structure for soluble proteins could not apply to transmembrane proteins directly for the same reason.Based on the comparison between transmembrane proteins and soluble proteins on their biochemical and biophysical properties, it is considered that the many methods for structure prediction of soluble proteins could be employed to solved the same issue of transmembrane proteins, but only when they are improved according to the specificity of transmembrane proteins, in which TMP-specific features and the corresponding improved algorithms are vitally required. By doing this, the specificity of transmembrane proteins is not the obstacle to predict their structure, but on the contrary benefits the issue. Therefore, we selected many TMP-specific features into the algorithms, and constructed the corresponding computational system for transmembrane protein fold recognition, in which folds can be recognized using the same method for both a transmembrane proteins and β transmembrane proteins.The contents of this studying are the following:1) The methods for soluble protein structure prediction have been deep studied, and the special structural properties of transmembrane proteins are analyzed and compared to that of soluble proteins. Then, an approach for fold recognition of transmembrane protein was designed and evaluated available, namely, the folds will be recognized step by step according to the order:amino acid sequenceâ†'topology structure predictionâ†'sequence-to-structure alignmentâ†'fold recognition.2) Many available and reliable TMP-specific features were found and extracted based on the corresponding researches and released programs in recent. These features were employed to offer more structure relative information for each studying step.3) For the reason that current methods cannot satisfy the requirement of topology prediction accuracy for a transmembrane proteins, we firstly introduced the inter-helical contact as structural feature to improve the topology prediction accuracy, and an adaptable support vector machine (SVM) model were designed for identification of transmembrane helixes. Correspondingly, a consensuses topology prediction method, CNTOP, were implemented and released.4) Segment type and transmembrane segment orientation were extracted from topology-based features and further applied to sequence-to-structure alignment methods TMSA and OMSA, respectively for a transmembrane proteins and β transmembrane proteins. Compared with peer methods for general protein (there is no such method for transmembrane proteins), our methods achieved better alignment accuracy, and the alignment score were proved having ability to present the structure similarity between the target protein and template proteins, which demonstrates that both methods could be used for fold recognition. Notably, OMSA can further determinate (3transmembrane proteins from non-transmembrane proteins.5) Based on the previous research on topology prediction and segment-based alignment methods, we developed an individual fold recognition method, TMFR, for both type of transmembrane proteins. The sequence-to-structure alignment method integrated in TMFR employed transmembrane residue solvent accessibility prediction additionally and thereby further improved the alignment accuracy. As consequent, TMFR recognized the transmembrane folds more accurate than peer methods, and can be considered available to select high quantity template proteins for transmembrane protein structure prediction. This method fills the blank of transmembrane protein fold recognition.6) Researches in this paper discussed the key problems in structural studies of transmembrane protein, and proposed a series of resolutions for the corresponding problems. It offered many available methodologies pushing forward the studying of transmembrane proteins, and raised practical suggestions for further works.This paper investigated distinguish specificity of transmembrane proteins and extracted corresponding features for the research of three major problems in transmembrane proteins: topology structure prediction, sequence-to-structure alignment and fold recognition. All the methods were tested having better performances compared with peer methods. The results indicate that transmembrane specificity structural features perform well to improve transmembrane fold recognition as expected when they are integrated in the adaptable algorithms. It provides new insights and methodologies solving corresponding researches, which can further push the studies on transmembrane protein structure prediction and function prediction forward.