Mohammad Nassef

Multistage feature selection approach for high-dimensional cancer data

Cancer is a serious disease that causes death worldwide. DNA methylation (DNAm) is an epigenetic mechanism, which controls the regulation of gene expression and is useful in early detection of cancer. The challenge with DNA methylation microarray datasets is the huge number of CpG sites compared to the number of samples. Recent research efforts attempted to reduce this high dimensionality by different feature selection techniques. This article proposes a multistage feature selection approach to select the optimal CpG sites from three different DNAm cancer datasets (breast, colon and lung). The proposed approach combines three different filter feature selection methods including Fisher Criterion, t-test and Area Under ROC Curve. In addition, as a wrapper feature selection, we apply genetic algorithms with Support Vector Machine Recursive Feature Elimination (SVM-RFE) as its fitness function, and SVM as its evaluator. Using the Incremental Feature Selection (IFS) strategy, subsets of 24, 13 and 27 optimal CpG sites are selected for the breast, colon and lung cancer datasets, respectively. By applying fivefold cross-validation on the training datasets, these subsets of optimal CpG sites showed perfect classification accuracies of 100, 100 and 97.67%, respectively. Moreover, the testing of the three independent cancer datasets by these final subsets resulted in accuracies 96.02, 98.81 and 94.51%, respectively. The experimental results demonstrated high classification performance and small optimal feature subsets. Consequently, the biological significance of the genes corresponding to these feature subsets is validated using enrichment analysis.

Fractional canny edge detection for biomedical applications

This paper presents a comparative study of edge detection algorithms based on integer and fractional order differentiation. A performance comparison of the two algorithms has been proposed. Then, a soft computing technique has been applied to both algorithms for better edge detection. From the simulations, it shows that better performance is obtained compared to the classical approach. The noise performances of those algorithms are analyzed upon the addition of random Gaussian noise, as well as the addition of salt and pepper noise. The performance has been compared to peak signal to noise ratio (PSNR). From results, it is obtained that fractional edge detection with the fuzzy system has better performance.

Lung Cancer Detection and Classification with 3D Convolutional Neural Network (3D-CNN)

This paper demonstrates a computer-aided diagnosis (CAD) system for lung cancer classification of CT scans with unmarked nodules, a dataset from the Kaggle Data Science Bowl, 2017. Thresholding was used as an initial segmentation approach to segment out lung tissue from the rest of the CT scan. Thresholding produced the next best lung segmentation. The initial approach was to directly feed the segmented CT scans into 3D CNNs for classification, but this proved to be inadequate. Instead, a modified U-Net trained on LUNA16 data (CT scans with labeled nodules) was used to first detect nodule candidates in the Kaggle CT scans. The U-Net nodule detection produced many false positives, so regions of CTs with segmented lungs where the most likely nodule candidates were located as determined by the U-Net output were fed into 3D Convolutional Neural Networks (CNNs) to ultimately classify the CT scan as positive or negative for lung cancer. The 3D CNNs produced a test set Accuracy of 86.6%. The performance of our CAD system outperforms the current CAD systems in literature which have several training and testing phases that each requires a lot of labeled data, while our CAD system has only three major phases (segmentation, nodule candidate detection, and malignancy classification), allowing more efficient training and detection and more generalizability to other cancers.

Computational determination of the effects of virulent Escherichia coli and salmonella bacteriophages on human gut

BACKGROUND AND OBJECTIVE Salmonella and Escherichia coli are different types of bacteria that cause food poisoning in humans. In the elderly, infants and people with chronic conditions, it is very dangerous if Salmonella or E. coli gets into the bloodstream and then they must be treated by phage therapy. Treating Salmonella and E. coli by phage therapy affects the gut flora. This research paper presents a system for detecting the effects of virulent E. coli and Salmonella bacteriophages on human gut. METHODS A method based on Domain-Domain Interactions (DDIs) model is implemented in the proposed system to determine the interactions between the proteins of human gut bacteria and the proteins of bacteriophages that infect virulent E. coli and Salmonella. The system helps gastroenterologists to realize the effect of injecting bacteriophages that infect virulent E. coli and Salmonella on the human gut. RESULTS By testing the system over Enterobacteria phage 933W, Enterobacteria phage VT2-Sa and Enterobacteria phage P22, it resulted in four interactions between the proteins of the bacteriophages that infect E. coli O157:H7, E. coli O104:H4 and Salmonella typhimurium and the proteins of human gut bacterium strains. CONCLUSION Several effects were detected such as: antibacterial activity against a number of bacterial species in human gut, regulation of cellular differentiation and organogenesis during gut, lung, and heart development, ammonia assimilation in bacteria, yeasts, and plants, energizing defense system and its function in the detoxification of lipopolysaccharide, and in the prevention of bacterial translocation in human gut.

An Algorithm for Browsing the Referentially-compressed Genomes

Genome resequencing produces enormous amount of data daily. Biologists need to frequently mine this data with the provided processing and storage resources. Therefore, it becomes very critical to professionally store this data in order to efficiently browse it in a frequent manner. Reference-based Compression algorithms (RbCs) showed significant genome compression results compared to the traditional text compression algorithms. By avoiding the complete decompression of the compressed genomes, they can be browsed by performing partial decompressions at specific regions, taking lower runtime and storage resources. This paper introduces the inCompressi algorithm that is designed and implemented to efficiently pick sequences from genomes, that are compressed by an existing Reference-based Compression algorithm (RbC), through partial decompressions. Moreover, inCompressi performs a more efficient complete genome decompression compared to the original decompression algorithm. The experimental results showed a significant reduction in both runtime and memory consumption compared to the original algorithm.

Computational Determination of the Effects of Bacteriophage Bacteriophage Interactions in Human Body

BACKGROUND Chronic diseases are becoming more serious and widely spreading and this carries a heavy burden on doctors to deal with such patients. Although many of these diseases can be treated by bacteriophages, the situation is significantly dangerous in patients having concomitant more than one chronic disease, where conflicts between phages used in treating these diseases are very closer to happen. METHOD This research paper presents a method to detecting the Bacteriophage-Bacteriophage Interaction. This method is implemented based on Domain-Domain Interactions model and it was used to infer Domain-Domain Interactions between the bacteriophages injected in the human body at the same time. RESULTS By testing the method over bacteriophages that are used to treat tuberculosis, salmonella and virulent E.coli, many interactions have been inferred and detected between these bacteriophages. Several effects were detected for the resulted interactions such as: playing a role in DNA repair such as nonhomologous end joining, playing a role in DNA replication, playing a role in the interaction between the immune system and the tumor cells and playing a role in the stiff man syndrome. We revised all patents relating to bacteriophage bacteriophage interactions and phage therapy. CONCLUSION The proposed method is developed to help doctors to realize the effect of simultaneously injecting different bacteriophages into the human body to treat different diseases.

Simulated Annealing with Levy Distribution for Fast Matrix Factorization-Based Collaborative Filtering

Matrix factorization is one of the best approaches for collaborative filtering, because of its high accuracy in presenting users and items latent factors. The main disadvantages of matrix factorization are its complexity, and being very hard to be parallelized, specially with very large matrices. In this paper, we introduce a new method for collaborative filtering based on Matrix Factorization by combining simulated annealing with levy distribution. By using this method, good solutions are achieved in acceptable time with low computations, compared to other methods like stochastic gradient descent, alternating least squares, and weighted non-negative matrix factorization.

An Enhanced Approach for Detection and Classification of Computed Tomography Lung Cancer

The paper presents approaches for nodule detection and extraction in axial lung computed tomography. The goal is to detect correctly pulmonary nodule to recognize and screen lung cancer patients. The pulmonary nodule detection is very challenging problem. The proposed model developed a hybrid efficient model based on affine-invariant representation and shape of segmented nodule. Due to large number of extracted features for all slices on patient, feature selection is an important step to select the most important feature for classification. We apply forward stepwise least squares regression that maximizes the Rsquared value, this criterion provides a fast preprocessing feature selection assessment for systems with huge volumes of features based on a linear models framework. Moreover, gradient boosting have been suggested to select the relevant features based on boosting approach. Classification of patients has been done by support vector machine. Kaggle DSB dataset is used to test the accuracy of our model. The results show major improvement in accuracy and the features are reduced.

Feature Selection and Extraction Framework for DNA Methylation in Cancer

Feature selection methods for cancer classification are aimed to overcome the high dimensionality of the biomedical data which is a challenging task. Most of the feature selection methods based on DNA methylation are time consuming during testing phase to identify the best pertinent features subset that are relevant to accurate prediction. However, the hybridization between feature selection and extraction methods will bring a method that is far fast than only feature selection method. This paper proposes a framework based on both novel feature selection methods that employ statistical variation, standard deviation and entropy, along with extraction methods to predict cancer using three new features, namely, Hypomethylation, Midmethylation and Hypermethylation. These new features represent the average methylation density of the corresponding three regions. The three features are extracted from the selected features based on the analysis of the methylation behavior. The effectiveness of the proposed framework is evaluated by the breast cancer classification accuracy. The results give 98.85% accuracy using only three features out of 485,577 features. This result proves the capability of the proposed approach for breast cancer diagnosis and confirms that feature selection and extraction methods are critical for practical implementation.

A Comparative Study of Feature Selection and Classification Techniques for High-Throughput DNA Methylation Data

The high dimensionality of data is a common problem in classification. In this work, a small number of significant features is investigated to classify data of two sample groups. Various feature selection and classification techniques are applied in a collection of four high-throughput DNA methylation microarray data sets. Using accuracy as a performance metric, the repeated 10-fold cross-validation strategy is implemented to evaluate the different proposed techniques. Combining the Signal to Noise Ratio (SNR) and Wilcoxon rank-sum test filter methods with Support Vector Machine-Recursive Feature Elimination (SVM-RFE) as an embedded method has resulted in a perfect performance. In addition, the linear classifiers showed excellent results compared to others classifiers when applied to such data sets.