Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userImproving Neural Protein-Protein Interaction Extraction with Knowledge Selection
184
0
0.0
(
0
)
Added by
Chengkun Lang
Publication date
2019
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Protein-protein interaction (PPI) extraction from published scientific literature provides additional support for precision medicine efforts. Meanwhile, knowledge bases (KBs) contain huge amounts of structured information of protein entities and their relations, which can be encoded in entity and relation embeddings to help PPI extraction. However, the prior knowledge of protein-protein pairs must be selectively used so that it is suitable for different contexts. This paper proposes a Knowledge Selection Model (KSM) to fuse the selected prior knowledge and context information for PPI extraction. Firstly, two Transformers encode the context sequence of a protein pair according to each protein embedding, respectively. Then, the two outputs are fed to a mutual attention to capture the important context features towards the protein pair. Next, the context features are used to distill the relation embedding by a knowledge selector. Finally, the selected relation embedding and the context features are concatenated for PPI extraction. Experiments on the BioCreative VI PPI dataset show that KSM achieves a new state-of-the-art performance (38.08% F1-score) by adding knowledge selection.
rate research
Read More
The determination of protein functions is one of the most challenging problems of the post-genomic era. The sequencing of entire genomes and the possibility to access genes co-expression patterns has moved the attention from the study of single proteins or small complexes to that of the entire proteome. In this context, the search for reliable methods for proteins function assignment is of uttermost importance. Previous approaches to deduce the unknown function of a class of proteins have exploited sequence similarities or clustering of co-regulated genes, phylogenetic profiles, protein-protein interactions, and protein complexes. We propose to assign functional classes to proteins from their network of physical interactions, by minimizing the number of interacting proteins with different categories. The function assignment is made on a global scale and depends on the entire connectivity pattern of the protein network. Multiple functional assignments are made possible as a consequence of the existence of multiple equivalent solutions. The method is applied to the yeast Saccharomices Cerevisiae protein-protein interaction network. Robustness is tested in presence of a high percentage of unclassified proteins and under deletion/insertion of interactions.
From the spectral plot of the (normalized) graph Laplacian, the essential qualitative properties of a network can be simultaneously deduced. Given a class of empirical networks, reconstruction schemes for elucidating the evolutionary dynamics leading to those particular data can then be developed. This method is exemplified for protein-protein interaction networks. Traces of their evolutionary history of duplication and divergence processes are identified. In particular, we can identify typical specific features that robustly distinguish protein-protein interaction networks from other classes of networks, in spite of possible statistical fluctuations of the underlying data.
Sucralose is a commonly employed artificial sweetener that appears to destabilize protein native structures. This is in direct contrast to the bio-preservative nature of its natural counterpart, sucrose, which enhances the stability of biomolecules against environmental stress. We have further explored the molecular interactions of sucralose as compared to sucrose to illuminate the origin of the differences in their bio-preservative efficacy. We show that the mode of interactions of sucralose and sucrose in bulk solution differ subtly using hydration dynamics measurement and computational simulation. Sucralose does not appear to disturb the native state of proteins for moderate concentrations (<0.2 M) at room temperature. However, as the concentration increases, or in the thermally stressed state, sucralose appears to differ in its interactions with protein leading to the reduction of native state stability. This difference in interaction appears weak. We explored the difference in the preferential exclusion model using time-resolved spectroscopic techniques and observed that both molecules appear to be effective reducers of bulk hydration dynamics. However, the chlorination of sucralose appears to slightly enhance the hydrophobicity of the molecule, which reduces the preferential exclusion of sucralose from the protein-water interface. The weak interaction of sucralose with hydrophobic pockets on the protein surface differs from the behavior of sucrose. We experimentally followed up upon the extent of this weak interaction using isothermal titration calorimetry (ITC) measurements. We propose this as a possible origin for the difference in their bio-preservative properties.
Recently, the structural controllability theory has been introduced to analyze the Protein-Protein Interaction (PPI) network. The indispensable nodes, which their removal increase the number of driver nodes to control the network, are found essential in PPI network. However, the PPI network is far from complete and there may exist many false-positive or false-negative interactions, which promotes us to question: are these indispensable nodes robust to structural change? Here we systematically investigate the robustness of indispensable nodes of PPI network by removing and adding possible interactions. We found that the indispensable nodes are sensitive to the structural change and very few edges can change the type of many indispensable nodes. The finding may promote our understanding to the control principle of PPI network.
Answer selection, which is involved in many natural language processing applications such as dialog systems and question answering (QA), is an important yet challenging task in practice, since conventional methods typically suffer from the issues of ignoring diverse real-world background knowledge. In this paper, we extensively investigate approaches to enhancing the answer selection model with external knowledge from knowledge graph (KG). First, we present a context-knowledge interaction learning framework, Knowledge-aware Neural Network (KNN), which learns the QA sentence representations by considering a tight interaction with the external knowledge from KG and the textual information. Then, we develop two kinds of knowledge-aware attention mechanism to summarize both the context-based and knowledge-based interactions between questions and answers. To handle the diversity and complexity of KG information, we further propose a Contextualized Knowledge-aware Attentive Neural Network (CKANN), which improves the knowledge representation learning with structure information via a customized Graph Convolutional Network (GCN) and comprehensively learns context-based and knowledge-based sentence representation via the multi-view knowledge-aware attention mechanism. We evaluate our method on four widely-used benchmark QA datasets, including WikiQA, TREC QA, InsuranceQA and Yahoo QA. Results verify the benefits of incorporating external knowledge from KG, and show the robust superiority and extensive applicability of our method.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
