Advances in machine learning for predicting protein functions / 生物工程学报

Proteins play a variety of functional roles in cellular activities and are indispensable for life. Understanding the functions of proteins is crucial in many fields such as medicine and drug development. In addition, the application of enzymes in green synthesis has been of great interest, but the high cost of obtaining specific functional enzymes as well as the variety of enzyme types and functions hamper their application. At present, the specific functions of proteins are mainly determined through tedious and time-consuming experimental characterization. With the rapid development of bioinformatics and sequencing technologies, the number of protein sequences that have been sequenced is much larger than those can be annotated, thus developing efficient methods for predicting protein functions becomes crucial. With the rapid development of computer technology, data-driven machine learning methods have become a promising solution to these challenges. This review provides an overview of protein function and its annotation methods as well as the development history and operation process of machine learning. In combination with the application of machine learning in the field of enzyme function prediction, we present an outlook on the future direction of efficient artificial intelligence-assisted protein function research.

Bovine TMEM95 gene: Polymorphisms detecting in five Chinese indigenous cattle breeds and their association with growth traits

BACKGROUND: Transmembrane protein 95 (TMEM95) plays a role in male fertility. Previous studies showed that genes with a significant impact on reproductive traits can also affect the growth traits of livestock. Thus, we speculated that the genetic variation of TMEM95 gene may have effects on growth traits of cattle. RESULTS: Two SNPs were genotyped. The rs136174626 and rs41904693 were in the intron 4 and 30 -untranslated region, respectively. The linkage disequilibrium analysis illustrated that these two loci were not linked. The rs136174626 was associated with six growth traits of Nanyang cattle, four traits of Luxi cattle, and three traits of Ji'an cattle. For rs41904693 locus, the GG individuals had greater body height and abdominal girth in Ji' an cattle than TT and TG individuals. In Jinnan cattle, GG and TT individuals had greater body height, height at hip cross, body length, and heart girth than TG individuals. The potential splice site prediction results suggest that the rs136174626 may influence the splicing efficiency of TMEM95, and the miRNA binding site prediction results showed that the rs41904693 may influence the expression of TMEM95 by affecting the binding efficiency of Bta-miR-1584 and TMEM95 30 -UTR. CONCLUSIONS: The findings of the study suggested that the two SNPs in TMEM95 could be a reliable basis for molecular breeding in cattle.

High-throughput sequencing analysis of the microbiota of subgingival plaque in patients with periodontitis / 口腔疾病防治

Objective @# To detect the composition of the subgingival microbiota in generalized aggressive periodontitis (GAgP) and severe chronic periodontitis (SCP) patients tested by high-throughput sequencing (HTS) technologies, analyze its diversity and function by using bioinformatics, and observe changes in the subgingival microbiota before and after periodontal initial therapy.@* Methods@#Eleven patients with GAgP and 14 patients with SCP who visited the Department of Periodontics in Stomatological Hospital of Kunming Medical University from September 2018 to May 2019 were recruited, and subgingival plaque samples were collected at baseline and 6 weeks after initial therapy. Then, the genomic DNA was distracted and sequenced by the Illumina MiSeq high-throughput sequencing platform. QIIME (quantitative insights in microbial ecology), Mothur, SPSS and other software were used to analyze community information. LEfSe difference analysis (linear discriminant analysis effect size), network analysis, and the KEGG PATHWAY database (https://www.kegg.jp/kegg/pathway.html) were used to predict community function. @* Results @# At baseline, the dominant microbiota of GAgP and SCP patients were similar, including Bacteroidetes, Porphyromonas and Porphyromonas endodontalis. Six weeks after initial therapy, as the periodontal pocket became shallower, the variation trend of the microbiota of GAgP and SCP patients was similar. The relative abundance of gram-negative bacteria, such as Bacteroidetes, Porphyromonas and Porphyromonas endodontalis, decreased, while the relative abundance of gram-positive bacteria, such as Proteobacteria, Actinomyces and Rothia aeria, increased. Actinobacteria were significantly increased biomarkers of the subgingival microbiota in GAgP after treatment. Streptococcus is an important genus that connects the microbiota related to periodontitis and the microbiota related to periodontal health. Community function prediction result showed that initial treatment can reduce the functions of amino acid metabolism, methane metabolism, and peptidase in GAgP and SCP patients.@*Conclusion@#The subgingival microbiota of GAgP and SCP patients are similar. Streptococcus, as an early colonizer, may play an important role in promoting plaque biofilm formation and maturation in the process of subgingival flora from health to imbalance. Initial therapy can change the composition and structure of the subgingival microbiota, reduce community diversity, and reduce the functions of amino acid metabolism, methane metabolism, and peptidase in GAgP and SCP patients.

In Silico Characterization of Meloidogyne Genus Nematode Cellulose Binding Proteins

Abstract Root-knot nematodes are a group of endoparasites species that induce the formation of giant cells in the hosts, by which they guarantee their feeding and development. Meloidogyne species infect over 2000 plant species, and are highly destructive, causing damage to many crops around the world. M. enterolobii is considered the most aggressive species in tropical regions, such as Africa and South America. Phytonematodes are able to penetrate and migrate within plant tissues, establishing a sophisticated interaction with their hosts through parasitism factors, which include a series of cell wall degradation enzymes and plant cell modification. Among the parasitism factors documented in the M. enterolobii species, cellulose binding protein (CBP), a nematode excretion protein that appears to be associated with the breakdown of cellulose present in the plant cell wall. In silico analysis can be of great importance for the identification, structural and functional characterization of genomic sequences, besides making possible the prediction of structures and functions of proteins. The present work characterized 12 sequences of the CBP protein of nematodes of the genus Meloidogyne present in genomic databases. The results showed that all CBP sequences had signal peptide and that, after their removal, they had an isoelectric point that characterized them as unstable in an acid medium. The values of the average hydrophilicity demonstrated the hydrophilic character of the analyzed sequences. Phylogenetic analyzes were also consistent with the taxonomic classification of the nematode species of this study. Five motifs were identified, which are present in all sequences analyzed. These results may provide theoretical grounds for future studies of plant resistance to nematode infection.

Estimación de la función renal mediante creatinina sérica y fórmulas predictivas en población litiásica cubana / Stimation of renal function from serum creatinine and predictive formulas in Cuban population with lithiasic

Introducción: la función renal puede ser estimada mediante la creatinina sérica o por fórmulas predictivas, de ahí que resulte importante evidenciar el valor de estos métodos. Objetivo: determinar la validez de la creatinina sérica y la fiabilidad de las fórmulas de estimación de la función renal en población litiásica cubana. Métodos: se realizó un estudio transversal. Se estudiaron 6 290 pacientes con litiasis urinarias que se realizaron estudio metabólico en el Instituto de Nefrología entre 2006 y 2011, 4 133 (65,7 por ciento) del sexo masculino y 2 157 (34,3 por ciento), del femenino. La información fue procesada utilizando el paquete estadístico SPSS 15.0. Para evaluar la capacidad diagnóstica de la creatinina sérica se emplearon curvas ROC y en el análisis de la fiabilidad de las fórmulas, diagramas de Bland y Altman. Resultados: las fórmulas con menor diferencia promedio con respecto al aclaramiento de creatinina fueron en la ERC (TFG ≤ 90 mL/min/1,73 m²): Cockcroft-Gault (3,81 mL/min/1,73 m²; DE 19,20 mL/min/1,73 m²), Salazar-Corcoran (-4,47 mL/min/1,73 m²; DE 18,40 mL/min/1,73 m²) y Levey (MDRD) (4,69 mL/min/1,73 m²; DE 13,75 mL/min/1,73 m²) y en IRC (TFG < 60 mL/min/1,73 m ²): Levey (MDRD) (1,39 mL/min/1,73 m²; DE 10,22 mL/min/1,73 m²), Jelliffe 1 973 (2,15 mL/min/1,73 m²; DE 10,87 mL/min/1,73 m²) y Salazar-Corcoran (-4,47 mL/min/1,73 m²; DE 18,40 mL/min/1,73 m²). La ecuación de Baracskay tuvo la mayor diferencia promedio (-9,67 mL/min/1,73m²; DE 17,04 mL/min/1,73 m²). El valor óptimo de corte para la creatinina sérica en ERC fue 1,07 mg/dLy 0,89 mg/dL, en hombres y mujeres, respectivamente. Conclusiones: los resultados de este estudio sugieren que la fiabilidad de las fórmulas predictivas es alta, con excepción de la de Baracskay, utilizada en ancianos...

Introduction: renal function may be estimated from serum creatinine or with prediction formulas. Hence the importance of being aware of the usefulness of these methods. Objective: determine the validity of serum creatinine and the reliability of renal function estimation formulas. Methods: a cross-sectional study was conducted of 6 290 patients with urolithiasis (4 133 male and 2 157 female -65.7% and 34.3 percent, respectively-) undergoing metabolic studies at the Institute of Nephrology from 2006 to 2011. Data were processed with the statistical software SPSS version 15.0. ROC curves were used to evaluate the diagnostic capacity of serum creatinine. Reliability of the formulas was assessed with Bland-Altman plots. Results: the formulas with the smallest mean difference with respect to creatinine clearance were the following: for ERC (GFR ≤ 90 mL/min/1.73 m²): Cockcroft-Gault (3.81 mL/min/1.73 m²; DE 19.20 mL/min/1.73 m²), Salazar-Corcoran (-4.47 mL/min/1.73 m²; DE 18.40 mL/min/1.73 m²) and Levey (MDRD) (4.69 mL/min/1.73 m²; DE 13.75 mL/min/1.73 m²) and for IRC (GFR < 60 mL/min/1.73 m ²): Levey (MDRD) (1.39 mL/min/1.73 m²; DE 10.22 mL/min/1.73 m²), Jelliffe 1 973 (2.15 mL/min/1.73 m²; DE 10.87 mL/min/1.73 m²) and Salazar-Corcoran (-4.47 mL/min/1.73 m²; DE 18.40 mL/min/1.73 m²). The Baracskay formula showed the greatest mean difference (-9.67 mL/min/1.73m²; DE 17.04 mL/min/1.73 m²). Optimal cutoff value for serum creatinine in ERC was 1.07 mg/dL and 0.89 mg/dL for men and women, respectively. Conclusions: results suggest that the reliability of prediction formulas is high, except for Baracskay's, which is used in elderly patients...

Review of Biological Network Data and Its Applications

Studying biological networks, such as protein-protein interactions, is key to understanding complex biological activities. Various types of large-scale biological datasets have been collected and analyzed with high-throughput technologies, including DNA microarray, next-generation sequencing, and the two-hybrid screening system, for this purpose. In this review, we focus on network-based approaches that help in understanding biological systems and identifying biological functions. Accordingly, this paper covers two major topics in network biology: reconstruction of gene regulatory networks and network-based applications, including protein function prediction, disease gene prioritization, and network-based genome-wide association study.

Predicting enzyme class from protein structure using Bayesian classification

Predicting enzyme class from protein structure parameters is a challenging problem in protein analysis. We developed a method to predict enzyme class that combines the strengths of statistical and data-mining methods. This method has a strong mathematical foundation and is simple to implement, achieving an accuracy of 45%. A comparison with the methods found in the literature designed to predict enzyme class showed that our method outperforms the existing methods.