DNA Splice Research Articles

A hybrid approach of ensemble learning and grey wolf optimizer for DNA splice junction prediction.

DNA splice junction classification is a crucial job in computational biology. The challenge is to predict the junction type (IE, EI, or N) from a given DNA sequence. Predicting junction type is crucial for understanding gene expression patterns, disease causes, splicing regulation, and gene structure. The location of the regions where exons are joined, and introns are removed during RNA splicing is very difficult to determine because no universal rule guides this process. This study presents a two-layer hybrid approach inspired by ensemble learning to overcome this challenge. The first layer applies the grey wolf optimizer (GWO) for feature selection. GWO's exploration ability allows it to efficiently search a vast feature space, while its exploitation ability refines promising areas, thus leading to a more reliable feature selection. The selected features are then fed into the second layer, which employs a classification model trained on the retrieved features. Using cross-validation, the proposed method divides the DNA splice junction dataset into training and test sets, allowing for a thorough examination of the classifier's generalization ability. The ensemble model is trained on various partitions of the training set and tested on the remaining held-out fold. This process is performed for each fold, comprehensively evaluating the classifier's performance. We tested our method using the StatLog DNA dataset. Compared to various machine learning models for DNA splice junction prediction, the proposed GWO+SVM ensemble method achieved an accuracy of 96%. This finding suggests that the proposed ensemble hybrid approach is promising for DNA splice junction classification. The implementation code for the proposed approach is available at https://github.com/EFHamouda/DNA-splice-junction-prediction.

Predicting DNA sequence splice site based on graph convolutional network and DNA graph construction

Identifying splice sites is essential for gene structure analysis and eukaryotic genome annotation. Recently, computational and deep learning approaches for splice site detection have advanced, focusing on reducing false positives by distinguishing true from pseudo splice sites. This paper introduces GraphSplice, a method using graph convolutional neural networks. It encodes DNA sequences into directed graphs to extract features and predict splice sites. Tested across multiple datasets, GraphSplice consistently achieved high accuracy (91%-94%) and F1Scores (92%-94%), outperforming state-of-the-art models by up to 9.16% for donors and 5.64% for acceptors. Cross-species experiments also show GraphSplice’s capability to annotate splice sites in under-trained genomic datasets, proving its wide applicability as a tool for DNA splice site analysis.

CNSC-06. TEMOZOLOMIDE-INDUCED GUANINE MUTATIONS CREATE EXPLOITABLE VULNERABILITIES OF GUANINE-RICH DNA AND RNA REGIONS IN DRUG-RESISTANT GLIOMAS

Abstract Temozolomide (TMZ) is a chemotherapeutic agent that has been the first-line standard of care for the aggressive braincancer glioblastoma (GBM) since 2005. Although initially beneficial, TMZ resistance is universal and second-lineinterventions are an unmet clinical need. Here, we took advantage of the known mechanism of action of TMZ to targetguanines (G) and investigated G-rich G-quadruplex (G4) and splice site changes that occur upon TMZ resistance. We reportthat TMZ-resistant GBM has guanine mutations that disrupt the G-rich DNA G4s and splice sites that lead to deregulatedalternative splicing. These alterations create vulnerabilities, which are selectively targeted by either the G4-stabilizingdrug TMPyP4 or a novel splicing kinase inhibitor of cdc2-like kinase. Lastly, we show that the G4 and RNA binding proteinEWSR1, member of the FUS, EWSR1, and TAF15 family, aggregates in the cytoplasm in TMZ-resistant GBM cells and patient samples. Together, our findings provide insight into targetable vulnerabilities of TMZ-resistant GBM and present cytoplasmic EWSR1 as a putative biomarker.

Temozolomide-induced guanine mutations create exploitable vulnerabilities of guanine-rich DNA and RNA regions in drug-resistant gliomas.

Temozolomide (TMZ) is a chemotherapeutic agent that has been the first-line standard of care for the aggressive brain cancer glioblastoma (GBM) since 2005. Although initially beneficial, TMZ resistance is universal and second-line interventions are an unmet clinical need. Here, we took advantage of the known mechanism of action of TMZ to target guanines (G) and investigated G-rich G-quadruplex (G4) and splice site changes that occur upon TMZ resistance. We report that TMZ-resistant GBM has guanine mutations that disrupt the G-rich DNA G4s and splice sites that lead to deregulated alternative splicing. These alterations create vulnerabilities, which are selectively targeted by either the G4-stabilizing drug TMPyP4 or a novel splicing kinase inhibitor of cdc2-like kinase. Last, we show that the G4 and RNA binding protein EWSR1 aggregates in the cytoplasm in TMZ-resistant GBM cells and patient samples. Together, our findings provide insight into targetable vulnerabilities of TMZ-resistant GBM and present cytoplasmic EWSR1 as a putative biomarker.

When should we order a next generation sequencing test in a patient with cancer?

Technical advances in genome sequencing and the implementation of next-generation sequencing (NGS) in clinical oncology have paved the way for individualizing cancer patient therapy based on molecular profiles. When and how to use NGS testing in the clinic is at present an unsolved issue, although new research results provide evidence favoring this approach in some types of advanced cancer. Clinical research is evolving rapidly, from basket and umbrella trials to adaptative design precision oncology clinical studies, and genomic and molecular data often displace the classical clinical validation procedures of biomarkers. In this context, physicians must be aware of the clinical evidence behind these new biomarkers and NGS tests available, in order to use them in the right moment, and with a critical point of view. This review will present the status of currently available targeted drugs that can be effective based on actionable molecular alterations, and the NGS tests that are currently available, offering a practical guide for the application of Clinical Precision Oncology in the real world routine practice.

Abstract 3740: The pursue of genetic mechanisms underlying supreme response to pazopanib treatment

Abstract Aim: Studying exceptional responders to standard treatments may provide broader insights into the genetic mechanisms underlying such responses and suggest therapeutic options that weren’t previously apparent. In this study we unravelled the genetic signature of undifferentiated sarcoma in order to disclose the molecular mechanisms that may underlie the patient’s exquisite sensitivity to Pazopanib. Methods: A female patient with massive undifferentiated gynecological sarcoma experienced a dramatic response to Pazopanib that made debulking surgery possible. We performed whole exome sequencing on DNA isolated from both blood and formalin fixed paraffin-embedded (FFPE) tumour employing Ion Torrent technology. Variant annotation was performed using the Ion ReporterTM v5.0 (MA, USA) tool after application of specific filters. Subsequent curation of annotated data was completed using a combined semi-automated data analyses workflow developed by Minotech Genomics Facility with the implementation of two in house-developed lists regarding 24 pazopanib-related genes, and 96 angiogenesis-associated genes. We also performed targeted sequencing of Ion Ampliseq Comprehensive Cancer panel (CCP) that is strategically designed to interrogate coding DNA sequences and splice variants across 409 tumour suppressor genes and oncogenes. Results: Whole exome sequencing was conducted at 20x coverage for 91.9% and 60.5% of >19,000 genes for germline and FFPE DNA samples, respectively. Targeted sequencing was conducted at 20x for >97% of 409 genes of the comprehensive cancer panel. Comparative analyses revealed two variants common in germline and FFPE samples, localized in LTK(p.Val541Ile) and FGFR4 (p.Gly388Arg). Interestingly, examination of the annotated data for biologically plausible mechanisms revealed 2 yet uncharacterized heterozygous variants in two other members of the FGFR family, FGFR2 (c.2183A>G, p.Asn728Ser) and FGFR3 (c.2050G>A, p.Val684Ile) that were present only in the FFPE samples. Even though both variants reside within the catalytic tyrosine kinase domains, they are predicted to exert tolerated effects on the encoded FGFR2 and FGFR3 proteins. Since, Asn728Ser is highly conserved among different kinases of the FGFR family we chose this variant to predict a possible effect on Pazopanib binding. Specifically, protein structure simulation of FGFR2 revealed that asparagine residue at position 728 is located in the external side of the drug binding cavity and is thus predicted to not affect Pazopanib binding. Conclusion: We pursue the possible genetic or somatic modifiers underling the excessive response to Pazopanib. The detection of two, yet uncharacterised, variants in known Pazopanib targets highlights the potential of next generation sequencing in disclosing the genetic signature of undifferentiated sarcomas in relation to the activity and effectiveness of certain therapeutic agents. Note: This abstract was not presented at the meeting. Citation Format: Emmanouil Saloustros, Helen Latsoudis, Despoina Vassou, Galateia Kallergi, Irene Stratidaki, Niki-Antonia Gounalaki, Vasiliki Fadouloglou, Ioannis Drositis, Emmanouil Kontopodis, Kitty Pavlakis, Dimitrios Mavroudis, Nikolaos Androulakis, Dimitrios Kafetzopoulos. The pursue of genetic mechanisms underlying supreme response to pazopanib treatment [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 3740. doi:10.1158/1538-7445.AM2017-3740

A novel approach for predicting DNA splice junctions using hybrid machine learning algorithms

Accurate identification of splice junctions in a DNA sequence is an active area of research. The knowledge of splice junction's occurrence provides valuable information about its internal genomic structure and aids in its deeper analysis and interpretation. The major problems faced during gene analysis are diversity, complexity and the uncertainty nature of DNA sequences. The application of computational techniques using machine learning algorithms in this direction has attracted enormous attention in the last few decades. In this study, the development of hybrid machine learning ensembles approaches is presented that address the splice junction problem more effectively. Multiple classifier systems consisting of random subspace, rotation forest and boosting methods are implemented and are validated over the real genome sequence dataset. A novel feature selection technique based on attribute's correlation estimation using Best first strategy is proposed. The average prediction accuracy achieved is more than 98 % in identifying the splice junctions. All the computations are performed with 95 % confidence interval. The results presented in this study are superior as compared to the state-of-the-art approaches in the literature. This work strengthens the viability of expanding and using machine learning models to similar problems.

Erratum to: Two new methods for DNA splice site prediction based on neuro-fuzzy network and clustering

Erratum to: Two new methods for DNA splice site prediction based on neuro-fuzzy network and clustering

Two new methods for DNA splice site prediction based on neuro-fuzzy network and clustering

Nowadays, genetic disorders, like cancer and birth defects, are a great threat to human life. Since the first noticing of these types of diseases, many efforts have been made and researches performed in order to recognize them and find a cure for them. These disorders affect genes and they appear as abnormal traits in a genetic organism. In order to recognize abnormal genes, we need to predict splice sites in a DNA signal; then, we can process the genetic codes between two continuous splice sites and analyze the trait that it represents. In addition to abnormal genes and their consequent disorders, we can also identify other normal human traits like physical and mental features. So the primary issue here is to estimate splice sites precisely. In this paper, we have introduced two new methods in using neuro-fuzzy network and clustering for DNA splice site prediction. In this method, instead of using raw data and nucleotide sequence as an input to neural network, a survey on the first bunch of the nucleotide sequence of true and false categories of the input is carried out and training of the neuro-fuzzy network is achieved based on the similarities and dissimilarities of the selected sequences. In addition, sequences of the large input data are clustered into smaller categories to improve the prediction as they are really spliced based on different mechanisms. Experimental results show that these improvements have increased the recognition rate of the splice sites.

An Evolutionary Algorithm Approach for Feature Generation from Sequence Data and Its Application to DNA Splice Site Prediction

Associating functional information with biological sequences remains a challenge for machine learning methods. The performance of these methods often depends on deriving predictive features from the sequences sought to be classified. Feature generation is a difficult problem, as the connection between the sequence features and the sought property is not known a priori. It is often the task of domain experts or exhaustive feature enumeration techniques to generate a few features whose predictive power is then tested in the context of classification. This paper proposes an evolutionary algorithm to effectively explore a large feature space and generate predictive features from sequence data. The effectiveness of the algorithm is demonstrated on an important component of the gene-finding problem, DNA splice site prediction. This application is chosen due to the complexity of the features needed to obtain high classification accuracy and precision. Our results test the effectiveness of the obtained features in the context of classification by Support Vector Machines and show significant improvement in accuracy and precision over state-of-the-art approaches.

Hepatitis B Virus DNA Splicing in Lebanese Blood Donors and Genotype A to E Strains: Implications for Hepatitis B Virus DNA Quantification and Infectivity

Hepatitis B virus (HBV) is one of the major viruses transmissible by blood that causes chronic infection in immunocompromised individuals. The study of 61 HBV carrier blood donors from Lebanon revealed multiple patterns of spliced HBV DNA. HBV DNA splicing was examined and quantified in samples of five genotypes and in seroconversion panels. The Lebanese sample median viral load was 1.5 ×10(2) IU/ml. All strains were genotype D, serotype ayw; 35 clustered as subgenotype D1 and 7 clustered as subgenotype D2. Three splice variants (SP1, SP1A, and Pol/S) were observed in 12 high-viral-load samples. Twenty samples of each genotype, A to E, were tested for the presence of HBV spliced DNA and SP1-specific splice variant. An unspliced HBV genome was dominant, but 100% of strains with a viral load of ≥10(5) copies/ml contained various proportions of spliced DNA. SP1 was detected in 56/100 (56%) samples in levels that correlated with the overall viral load. HBV DNA quantification with S (unspliced) and X (total DNA) regions provided different levels of viral load, with the difference corresponding to spliced DNA. During the highly infectious window period, the SP1 variant became detectable shortly after the hepatitis B surface antigen (HBsAg), suggesting a correlation between the initiation of splicing and the production of detectable levels of HBsAg. The quantification of HBV DNA with primers located outside and inside the spliced region might provide different estimations of viral load and differentiate between infectious and defective viral genomes. The role of splicing neoproteins in HBV replication and interaction with the host remains to be determined.

Expression of Linear and Novel Circular Forms of an INK4/ARF-Associated Non-Coding RNA Correlates with Atherosclerosis Risk

Human genome-wide association studies have linked single nucleotide polymorphisms (SNPs) on chromosome 9p21.3 near the INK4/ARF (CDKN2a/b) locus with susceptibility to atherosclerotic vascular disease (ASVD). Although this locus encodes three well-characterized tumor suppressors, p16INK4a, p15INK4b, and ARF, the SNPs most strongly associated with ASVD are ∼120 kb from the nearest coding gene within a long non-coding RNA (ncRNA) known as ANRIL (CDKN2BAS). While individuals homozygous for the atherosclerotic risk allele show decreased expression of ANRIL and the coding INK4/ARF transcripts, the mechanism by which such distant genetic variants influence INK4/ARF expression is unknown. Here, using rapid amplification of cDNA ends (RACE) and analysis of next-generation RNA sequencing datasets, we determined the structure and abundance of multiple ANRIL species. Each of these species was present at very low copy numbers in primary and cultured cells; however, only the expression of ANRIL isoforms containing exons proximal to the INK4/ARF locus correlated with the ASVD risk alleles. Surprisingly, RACE also identified transcripts containing non-colinear ANRIL exonic sequences, whose expression also correlated with genotype and INK4/ARF expression. These non-polyadenylated RNAs resisted RNAse R digestion and could be PCR amplified using outward-facing primers, suggesting they represent circular RNA structures that could arise from by-products of mRNA splicing. Next-generation DNA sequencing and splice prediction algorithms identified polymorphisms within the ASVD risk interval that may regulate ANRIL splicing and circular ANRIL (cANRIL) production. These results identify novel circular RNA products emanating from the ANRIL locus and suggest causal variants at 9p21.3 regulate INK4/ARF expression and ASVD risk by modulating ANRIL expression and/or structure.

DNA splice site sequences clustering method for conservativeness analysis

DNA sequences that are near to splice sites have remarkable conservativeness, and many researchers have contributed to the prediction of splice site. In order to mine the underlying biological knowledge, we analyze the conservativeness of DNA splice site adjacent sequences by clustering. Firstly, we propose a kind of DNA splice site sequences clustering method which is based on DBSCAN, and use four kinds of dissimilarity calculating methods. Then, we analyze the conservative feature of the clustering results and the experimental data set.

Identifying DNA splice sites using hypernetworks with artificial molecular evolution

Identifying DNA splice sites is a main task of gene hunting. We introduce the hyper-network architecture as a novel method for finding DNA splice sites. The hypernetwork architecture is a biologically inspired information processing system composed of networks of molecules forming cells, and a number of cells forming a tissue or organism. Its learning is based on molecular evolution. DNA examples taken from GenBank were translated into binary strings and fed into a hypernetwork for training. We performed experiments to explore the generalization performance of hypernetwork learning in this data set by two-fold cross validation. The hypernetwork generalization performance was comparable to well known classification algorithms. With the best hypernetwork obtained, including local information and heuristic rules, we built a system (HyperExon) to obtain splice site candidates. The HyperExon system outperformed leading splice recognition systems in the list of sequences tested.

Breast Tumor Susceptibility to Chemotherapy Via Support Vector Machines

Support vector machine (SVM) classification together with alternative DNA splicing techniques were used to generate a classifier for breast cancer patients that are partial-responders to chemotherapy treatment. Partial responders are patients whose tumors were reduced by at least 50%. A stand-alone linear-programming-based SVM algorithm was used to separate the partial-responders from the nonresponders. A novel aspect of the classification approach utilized here is that each patient is represented by multiple points (replicates) in the 25-dimensional input space of DNA splice measurements.Replicates for all patients except those for one patient, were used as a training set. The average of the replicates for the patient left out was then used to test the leave one out correctness (looc). The looc for a group of 35 patients, with 9 partial-responders and 26 nonresponders was 88.6%, in an input space of 5 gene expressions extracted from an original space of 25 gene expression transcripts.

Splice site prediction using support vector machines with a Bayes kernel

One of the most important tasks in correctly annotating genes in higher organisms is to accurately locate the DNA splice sites. Although relatively high accuracy has been achieved by existing methods, most of these prediction methods are computationally extensive. Due to the enormous amount of DNA sequences to be processed, the computational speed is an important issue to consider. In this paper, we present a new machine learning method for predicting DNA splice sites, which first applies a Bayes feature mapping (kernel) to project the data into a new feature space and then uses a linear Support Vector Machine (SVM) as a classifier to recognize the true splice sites. The computation time is linear to the number of sequences tested, while the performance is notably improved compared with the Naive Bayes classifier in terms of classification accuracy, precision, and recall. Our classification results are also comparable to the solution quality obtained by the SVMs with polynomial kernels, while the speed of our proposed method is significantly faster. This is a notable improvement in computational modeling considering the huge amount of DNA sequences to be processed.

Recognition of DNA Splice Junction via Machine Learning Approaches

Successful recognition of splice junction sites of human DNA sequences was achieved via three machine learning approaches. Both unsupervised (Kohonen’s Self-Organizing Map, KSOM) and supervised (Back-propagation Neural Network, BNN; and Support Vector Machine, SVM) machine learning techniques were used for the classification of sequences from the testing set into one of three categories: transition from exon to intron, transition from intron to exon, and no transition. The dataset used in this study is comprised of 1,424 DNA sequences obtained from the National Center for Bioinformatics Information (NCBI). Performance of the machine learning approaches were assessed by the construction of learning models from 1,000 sequences of the training set and evaluated on the 424 sequences of the testing set that is unknown to the learning model. Each sequence is a window of 32 nucleotides long with regions comprising -15 to +15 nucleotides from the dinucleotide splice site. Since the nucleotides (A, C, G, and T) are represented by four digit binary code (e.g. 0001, 0010, 0100, and 1000) the number of descriptors increased from 32 to 128. The performance of machine learning techniques in order of increasing accuracy are as follows SVM > BNN > KSOM, suggesting that SVM is a robust method in the identification of unknown splice site. Although KSOM gave lower prediction accuracy than the two supervised methods, it is fascinating that it was able to make such prediction based only on knowledge of the input whereas the supervised method requires that the output be known during training. It is expected that the Support Vector Machine method can provide a powerful computational tool for predicting the splice junction sites of uncharacterized DNA.

Cutting Edge: Predominant Expression of a Novel Ikaros Isoform in Normal Human Hemopoiesis

Murine studies implicate Ikaros proteins as regulators of hemopoiesis, particularly in the lymphoid lineages. High homology between murine and human Ikaros suggests that Ikaros expression in the two might be similar. However, initial human studies that focused on leukemia detected novel Ikaros transcripts in patient samples. Thus, novel Ikaros splice forms and DNA nonbinding isoforms were linked with malignancy. We undertook an extensive analysis of normal human Ikaros expression to determine whether novel mRNAs are expressed as proteins and the extent to which these splice variants are unique to leukemia. Here we show that both mRNA and protein for DNA nonbinding Ikaros isoforms and splice variants previously linked to leukemia are expressed in normal human cells. However, our studies identify a new Ikaros isoform not previously described in mouse or human. This isoform is the predominant Ikaros protein in normal human cells, but not in leukemia cell lines.

Calcium and Phospholipid Binding Properties of Synthetic .gamma.-Carboxyglutamic Acid-Containing Peptides with Sequence Counterparts in Human Protein C

Two peptides with counterpart sequences in the gamma-carboxyglutamic acid (Gla) domain of human protein C (PC) have been synthesized and characterized. One peptide contained 38 amino acids (38-mer) and spanned the region from the amino terminus of the protein to the DNA splice junction between the Gla domain and the following short helical stretch, and the second peptide (48-mer) included a 10 amino acid extension that has been designed to incorporate the exon for the helical segment that is thought to play a role in stabilizing the Ca(2+)-dependent conformation of the Gla domain of proteins of this class. The peptides were synthesized by solid-phase methodology, then oxidized to allow disulfide pairing, and finally purified by FPLC methodology. Chemical characterization showed that each peptide contained its full complement of Gla residues. Two types of Ca(2+)-binding sites were found in these peptides, tighter sights (2-3) with Kd values of 60-370 microM and a weaker set of sites (7-10) with a range of Kd values from 0.8 to 3.1 mM. In general, the 48-mer interacted with Ca2+ more tightly than the 38-mer. As revealed by circular dichroism analysis, and by reactivity with monoclonal antibodies that recognize both the unfolded form of the Gla domain as well as the Ca(2+)-dependent conformation of this same domain, the 38-mer and 48-mer underwent the Ca(2+)-induced conformational changes characteristic of the intact protein. Both peptides displayed Ca(2+)-dependent binding to negatively charged phospholipid vesicles (PL).(ABSTRACT TRUNCATED AT 250 WORDS)