The number of new cases of infection of the Coronavirus disease, COVID-19, is alarming in many places in the world. In several world countries, including USA, the infection rates and daily cases numbers are fairly high; and there are even some spike increases in some USA states. Since the USA is experiencing the highest number of daily new cases in the world from May through July, the most important question is when we will witness an effective decline in the number of daily new cases? This paper follows a data-driven approach to induce the disease decline values from two country groups in the world where the disease declined already to less than 25% of its peak daily new cases. We apply these country groups' models to predict the decline to 25% of the US's peak. We compiled, examined, and analyzed pandemic data and statistics of two countries: g1: 42 countries, and g2: 14 countries. We utilize their data in the prediction of the decline timeline of the US. Group g2 consists of 14 countries having a similar number of cases per one million population. The majority of the models predict that the decline to 25% of the US's peak will be around the end of November to the first week of October. The results are significant and impressive as it is highly demanded to have clues and methods for the timeline prediction of this pandemic in the USA.

In bioinformatics, the classification of research documents in relation to their subject matter (specifically, disease types) is a very important task with many applications in the field. In terms of text classification, the total number of available disease documents is limited with Medline being the main source, containing a mostly abstract-only collection of research papers. However, abstracts are an important tool in gauging the specific disease in question, and giving a quick summary of the research. We introduce an effective technique based on feature learning for inducing feature weights appropriate for text classification with small dataset (in training and testing). We applied and evaluated the proposed technique in text classification task with medical abstracts representing disease texts. The conducted experiments with small data sets ranging from 40 to 500 documents (abstracts composed of around 300 words) produced excellent and promising results with respect to classification accuracy and AUC. In essence, the proposed technique gives leverage to the class distribution over each attribute rather than the attribute distribution over the two classes. Our technique consistently showed higher performance than conventional methods, and continued to improve with a decrease in the number of documents across all disease abstract classification experiments.

The study and analysis of functional annotations of human genes and gene variants are important for understanding functional genomics and gene-disease associations. This paper investigates the process of functional annotation of gene variants using the gene ontology. The paper also presents a method for annotating gene variants with biological process terms from the gene ontology. The presented method relies on identifying the enriched biological processes of the genes associated with the given genetic variants. With significant level of , the results found to be biologically significant. Furthermore, we found that certain ontology annotations e.g. {GO:0031325; positive regulation of cellular metabolic process} are more connected with mutations via certain genes compared with normal biological process terms in the gene ontology.

Research studies on pathogenic mutations are important and beneficial for understanding disease development, prognosis and gene-disease associations. Moreover, pathogenic mutations have harmful consequences which may lead to certain diseases or medical conditions. This paper presents a method for inducing the most significant and accurate functions for a given set of mutations having one common aspect, e.g. specific disease. The proposed method is based on the directed acyclic graph of the gene ontology to identify the most significant least common subsumers to be used for functionally annotating the mutations under investigation. We applied the method on a large sets of mutations. The reported results in this paper are encouraging and suggest that a mutation can have, and can be annotated with a function, e.g. biological process, from the gene ontology just like the genes, which contributes into a more complete understanding of mutation pathogenicity, mutation-gene-disease relationships, and disease mechanisms.

Multifunctional genes possess special significance due to their role in multiple biological and molecular activities in many organisms. In this paper, we study multifunctional genes in human with gene functional annotations in Gene Ontology, and we utilize the extensive information available for genes related to gene functionalities from the Molecular Function and Biological Process aspects of the Gene Ontology. Specifically, we examine the gene ontology annotations of human genes to assess the gene multifunctionality and their relationship with human diseases. We examined and analyzed all human genes having gene ontology annotations from GOA database and using the OMIM diseases. We combined all OMIM disease genes with the GO annotated human genes in relation to human diseases and the results of the relationships between highly annotated human genes and diseases are highly significant.

Gene Ontology (GO), one of the fastest growing bioinformatics projects, seeks to capture the most complete knowledge about gene related products. Over the past few years, many semantic similarity measures have been proposed based on various aspects of the GO, such as the corpus-based information content of the terms or distance between the terms. In this work, we first introduce the concept of structural information content (IC) and structural least common ancestor (LCA). We then use these concepts as part of a hybrid approach to propose a novel semantic similarity measure, known as Hybrid Structural Similarity (HSS) that depends only on the structure of the ontology and not on external annotation corpus. We have carried out extensive experimentation using the entire protein interaction database of S. cerevisiae (yeast) and compared HSS against other popular methods. The results show that in all except two cases, our method is able to outperform all methods by a significant margin. Further, our method is dynamic and responds dynamically to the term additions and updates in the ontology.

Gene Ontology is one of the largest bioinformatics project that seeks to consolidate knowledge about genes through annotation of terms to three ontologies. In this work, we present a technique to find association relationships in the annotation terms for the Saccharomyces cerevisiae (SGD) genome. We first present a normalization algorithm to ensure that the annotation terms have a similar level of specificity. Association rule mining algorithms are used to find significant and non-trivial association rules in these normalized datasets. Metrics such as support, confidence, and lift can be used to evaluate the strength of found rules. We conducted experiments on the entire SGD annotation dataset and here we present the top 10 strongest rules for each of the three ontologies. We verify the found rules using evidence from the biomedical literature. The presented method has a number of advantages - it relies only on the structure of the gene ontology, has minimal memory and storage requirements, and can be easily scaled for large genomes, such as the human genome. There are many applications of this technique, such as predicting the GO annotations for new genes or those that have not been studied extensively.

Protein sub cellular localization prediction is an important task with significant applications including the discovery of molecular functions of proteins. A number of prediction techniques have been developed in the past years based on protein sequence information. In this paper, we propose a new technique for predicting sub cellular localizations of proteins using improved features of proteins extracted from protein sequences. The method is effective in inducing the features from protein sequences in multiple localizations. We evaluated the method using six datasets of proteins from bacteria, gram-negative and gram-positive, plant, and non-plant proteins and compared the results with recent methods. The evaluation results with six protein localization datasets showed that the method is promising and competitive for predicting protein localizations. This method is fairly effective in extracting strong features from protein sequences which will have significant impact on great deal of research work that relies mainly on protein sequence information for discovering molecular functions of proteins, drug design, and disease-protein associations.

Most of the intelligent knowledge-based applications contain components for measuring semantic similarity between terms. Many of the existing semantic similarity measures that use ontology structure as their primary source cannot measure semantic similarity between terms and concepts using multiple ontologies. This research explores a new way to measure semantic similarity between biomedical concepts using multiple ontologies. We propose a new ontology-structure-based technique for measuring semantic similarity in single ontology and across multiple ontologies in the biomedical domain within the framework of unified medical language system (UMLS). The proposed measure is based on three features: 1) cross-modified path length between two concepts; 2) a new feature of common specificity of concepts in the ontology; and 3) local granularity of ontology clusters. The proposed technique was evaluated relative to human similarity scores and compared with other existing measures using two terminologies within UMLS framework: medical subject headings and systemized nomenclature of medicine clinical term. The experimental results validate the efficiency of the proposed technique in single and multiple ontologies, and demonstrate that our proposed measure achieves the best results of correlation with human scores in all experiments.

We propose a new method for measuring the semantic similarity of genes based on path length between their annotation terms in the gene ontology. Our method applies an exponential transfer function to the average path length between two genes to compute their similarity. The non-linear measure ensures that the semantic similarity decreases with distance and proves to be quite competitive when compared to other measures. The advantage of the proposed measure is its simplicity and ease of implementation which gives it a great appeal in this domain. The measure uses only one feature (path length) for computing the similarity between genes. For validation purposes, we computed the similarity of genes from the Saccharomyces genome database (SGD) taking part in various cellular pathways. We analyzed 152 pathways from SGD and compared our similarity results with two of the leading measures. The proposed measure proved to be very competitive in all cases and the clustering results showed that our method is able to surpass the leading methods in certain cases.

Gene ontology (GO) has fast become a dependable source for determining gene functions, gene similarity, and gene clustering. Furthermore, using GO and gene annotation databases, with semantic similarity measures, is now more acceptable in bioinformatics as means for gene functional analysis. In this paper, we compare four semantic similarity measures to compute the similarity between genes using GO annotations within a gene clustering application. The similarity measures we examined in this study rely on different information sources and techniques in computing the similarity between genes. For example, we selected two measures that are based on information-content, one measure is based on ontology-structure, and the fourth measure is based on annotation terms within the GO hierarchy. In the evaluation, we used five gene datasets from yeast genome, and we analyzed the results based on various clustering metrics and criteria.

The application of semantic similarity measures on gene data using Gene Ontology (GO) and gene annotation information is becoming more widely used and acceptable in the recent years in bioinformatics. The purpose of this application can range from gene similarity to gene clustering. In this paper, we investigate a simple measure for gene similarity that relies on the path length between the GO annotation terms of genes to determine the similarity between them. The similarity values computed by the proposed measure for a set of genes will then be used for clustering the genes. In the evaluation, we compared the proposed measure with two widely used information-theoretic similarity measures, Resnik and Lin, using three datasets of genes. The experimental results and analysis of clusters validated the effectiveness of the proposed path length measure.

Finding the similarity between biomedical terms and concepts is a very important task for biomedical information extraction and knowledge discovery. We propose and investigate the feasibility of using MEDLINE as standard corpus and MeSH ontology for measuring semantic similarity between concepts in the biomedical domain within UMLS framework. We adapted information-based semantic similarity measures from general English and applied them into the biomedical domain to measure the similarity between biomedical terms. The experimental results show that, by using MEDLINE and MeSH ontology, the information-based similarity measures perform very well and produce high correlations with human ratings. The similarity measure of Jiang & Conrath achieved 82% correlation with human similarity scores, and the average correlation with human scores of three measures is approaching 78%. These results confirm that MEDLINE is an effective information source for measuring semantic similarity between biomedical terms and concepts

The existing volumes of biomedical texts available online drive the increasing need for automated techniques to analyze and extract knowledge from these information repositories. Recognizing and classifying biomedical terms in these texts is an important step for developing efficient techniques for knowledge discovery and information extraction from the literature. This paper presents a new technique for biomedical term classification in biomedical texts. The method is based on combing successful feature selection techniques {MI, X2) with machine learning (SVM) for biomedical term classification. We utilize the advances in feature selection techniques in IR and use them to select the key features for term identification and classification. We evaluated the method using Genia 3.0 corpus with about 3,000 to more than 34,000 biomedical term instances. The technique is effective, achieving impressive accuracy, precision, and recall; and with F-score approaching -90%, the method is superior or very competitive with the recently published results.

Text categorization is continuing to be one of the most researched NLP problems due to the ever-increasing amounts of electronic documents and digital libraries. In this paper, we present a new text categorization method that combines the distributional clustering of words and a learning logic technique, called Lsquare, for constructing text classifiers. The high dimensionality of text in a document has not been fruitful for the task of categorization, for which reason, feature clustering has been proven to be an ideal alternative to feature selection for reducing the dimensionality. We, therefore, use distributional clustering method (IB) to generate an efficient representation of documents and apply Lsquare for training text classifiers. The method was extensively tested and evaluated. The proposed method achieves higher or comparable classification accuracy and {\rm F}_1 results compared with SVM on exact experimental settings with a small number of training documents on three benchmark data sets WebKB, 20Newsgroup, and Reuters-21578. The results prove that the method is a good choice for applications with a limited amount of labeled training data. We also demonstrate the effect of changing training size on the classification performance of the learners.

This paper presents a new context-based method for automatic detection and extraction of similar and related words from texts. Finding similar words is a very important task for many NLP applications including anaphora resolution, document retrieval, text segmentation, and text summarization. Here we use word similarity to improve search quality for search engines in (general and) specific domains. Our method is based on rules for extracting the words in the neighborhood of a target word, then connecting this with the surroundings of other occurrences of the same word in the (training) text corpus. This is an on-going work, and is still under extensive testing. The preliminary results, however, are promising and encouraging more work in this direction.