Target Pairs Research Articles

Designing sensitive viral diagnostics with machine learning

Design of nucleic acid-based viral diagnostics typically follows heuristic rules and, to contend with viral variation, focuses on a genome’s conserved regions. A design process could, instead, directly optimize diagnostic effectiveness using a learned model of sensitivity for targets and their variants. Toward that goal, we screen 19,209 diagnostic–target pairs, concentrated on CRISPR-based diagnostics, and train a deep neural network to accurately predict diagnostic readout. We join this model with combinatorial optimization to maximize sensitivity over the full spectrum of a virus’s genomic variation. We introduce Activity-informed Design with All-inclusive Patrolling of Targets (ADAPT), a system for automated design, and use it to design diagnostics for 1,933 vertebrate-infecting viral species within 2 hours for most species and within 24 hours for all but three. We experimentally show that ADAPT’s designs are sensitive and specific to the lineage level and permit lower limits of detection, across a virus’s variation, than the outputs of standard design techniques. Our strategy could facilitate a proactive resource of assays for detecting pathogens.

Registered report: Moderators of the relationship between implicit and explicit measures of evaluation and identification

Researchers and theorists have explained variations in the relationship between implicit and explicit measures of social cognition by citing differences in measurement and cognitive processing. The present research explored how features of measured attributes and individual differences affect the relationship between implicit and explicit measures using existing data (Hussey, Hughes, & Nosek, 2018). In the original study, participants completed measures of explicit evaluation or identification (i.e., difference scores between self-report items) and implicit evaluation or identification (i.e., the Implicit Association Test; Greenwald, McGhee, & Schwartz, 1998) for one of 95 target pairs. First, we attempted to conceptually replicate Nosek's (2005) findings that self-presentation, distinctiveness, elaboration/importance, and complementarity moderated the relationship between implicit and explicit evaluations across individuals irrespective of attribute targets. These attribute features each moderated the relationship between implicit and explicit evaluations and accounted for unique variation in implicit-explicit relations. Second, we tested whether these features affected the relationship between implicit and explicit identifications. We found moderating effects of distinctiveness and complementarity but not of the other attribute features, likely due to their operationalization. Finally, we examined whether theoretically relevant individual difference measures moderated the relationship between implicit and explicit evaluations or identifications. Across almost all analyses, these measures (i.e., socially desirable responding, self-monitoring, need for cognition, personal need for structure, and need for cognitive closure) did not moderate implicit-explicit relations. Taken together, our results suggest that relationships between implicit and explicit measures reflect individual experiences with attribute targets rather than the cultural context, targets, or individuals themselves.

MACSima imaging cyclic staining (MICS) technology reveals combinatorial target pairs for CAR T\xa0cell\xa0treatment of solid tumors

Many critical advances in research utilize techniques that combine high-resolution with high-content characterization at the single cell level. We introduce the MICS (MACSima Imaging Cyclic Staining) technology, which enables the immunofluorescent imaging of hundreds of protein targets across a single specimen at subcellular resolution. MICS is based on cycles of staining, imaging, and erasure, using photobleaching of fluorescent labels of recombinant antibodies (REAfinity Antibodies), or release of antibodies (REAlease Antibodies) or their labels (REAdye_lease Antibodies). Multimarker analysis can identify potential targets for immune therapy against solid tumors. With MICS we analysed human glioblastoma, ovarian and pancreatic carcinoma, and 16 healthy tissues, identifying the pair EPCAM/THY1 as a potential target for chimeric antigen receptor (CAR) T cell therapy for ovarian carcinoma. Using an Adapter CAR T cell approach, we show selective killing of cells only if both markers are expressed. MICS represents a new high-content microscopy methodology widely applicable for personalized medicine.

DNA-encoded library versus RNA-encoded library selection enables design of an oncogenic noncoding RNA inhibitor

Nature evolves molecular interaction networks through persistent perturbation and selection, in stark contrast to drug discovery, which evaluates candidates one at a time by screening. Here, nature's highly parallel ligand-target search paradigm is recapitulated in a screen of a DNA-encoded library (DEL; 73,728 ligands) against a library of RNA structures (4,096 targets). In total, the screen evaluated ∼300 million interactions and identified numerous bona fide ligand-RNA three-dimensional fold target pairs. One of the discovered ligands bound a 5'GAG/3'CCC internal loop that is present in primary microRNA-27a (pri-miR-27a), the oncogenic precursor of microRNA-27a. The DEL-derived pri-miR-27a ligand was cell active, potently and selectively inhibiting pri-miR-27a processing to reprogram gene expression and halt an otherwise invasive phenotype in triple-negative breast cancer cells. By exploiting evolutionary principles at the earliest stages of drug discovery, it is possible to identify high-affinity and selective target-ligand interactions and predict engagements in cells that short circuit disease pathways in preclinical disease models.

Meta-Pixel-Driven Embeddable Discriminative Target and Background Dictionary Pair Learning for Hyperspectral Target Detection

In hyperspectral target detection, the spectral high-dimensionality, variability, and heterogeneity will pose great challenges to the accurate characterizations of the target and background. To alleviate the problems, we propose a Meta-pixel-driven Embeddable Discriminative target and background Dictionary Pair (MEDDP) learning model by combining low-dimensional embeddable subspace projection and the discriminative target and background dictionary pair learning. In MEDDP, the meta-pixel set is built by taking the merits of homogeneous superpixel segmentation and the local manifold affinity structures, which can significantly reduce the influence of spectral variability and find the most typical and informative prototype spectral signature. Afterward, an embeddable discriminative dictionary pair learning model is established to learn a target and background dictionary pair based on the structural incoherent constraint with embeddable subspace projection. The proposed joint learning strategy can reduce the high-dimensional redundant information and simultaneously enhance the discrimination and compactness of the target and background dictionaries. The proposed MEDDP model is solved by an iterative and alternate optimization algorithm and applied with the meta-pixel-level target detection method. Experimental results on four benchmark HSI datasets indicate that the proposed method can consistently yield promising performance in comparison with some state-of-the-art target detectors.

Enhancer RNA LINC00242-Induced Expression of PHF10 Drives a Better Prognosis in Pancreatic Adenocarcinoma.

Enhancer RNA is a kind of non-coding RNA, which is transcribed from the enhancer region of gene and plays an important role in gene transcription regulation. However, the role of eRNA in pancreatic adenocarcinoma (PAAD) is still unclear. In this study, we identified the key eRNA and its target gene in PAAD. The transcriptome data and clinical information of pancreatic cancer were downloaded from the UCSC Xena platform. A total of 2,695 eRNAs and its target gene predicted by the PreSTIGE method were selected as candidate eRNA–target pairs. After survival analysis, we found that LINC00242 was the eRNA most related to patients’ survival, and correlation analysis further indicated that LINC00242 and its target gene PHF10 had a significant co-expression relationship. Downregulation of LINC00242 was significantly associated with unfavorable clinicopathological features. Based on pan-cancer analysis, we found that LINC00242 was associated with the survival of multiple cancers, and LINC00242 was co-expressed with its target genes in multiple cancer types. External experiments further demonstrated that PHF10 was the downstream target gene of LINC00242. After ssGSEA analysis, PAAD patients were classified as high, medium, and low immune cell infiltration clusters. Compared with the low and medium immune infiltration clusters, the expression level of PHF10 was significantly upregulated in the high immune infiltration clusters. After performing the CIBERSORT algorithm, we found that there was a significant difference in the abundance of immune infiltrating cells between the PHF10 high- and low-expression groups. Additionally, the web tool TIMER was used to detect the distribution and expression of PHF10 in pan-cancer.

Relative letter-position coding revisited

The notion that the brain achieves visual word recognition by encoding the relative positions of letters with open-bigram representations (e.g., ‘h-e’, ‘h-r’ and ‘e-r’ driving recognition of ‘her’) has been successful in accounting for many behaviors and phenomena. However, one characteristic of open-bigrams has remained unexplored: How is the activation of a bigram modulated by the distance between its constituents in the visual field? On the one hand, contiguous letters (e.g., ‘at’ in ‘father’) may allow for a clearer percept of the bigram. On the other hand, an increasing distance between letters (e.g., ‘ae’ in ‘father’) should create more certainty about their relative positions, which is precisely what the bigram is meant to convey. This matter was investigated with two experiments in which participants indicated whether target pairs of letters occurred in random letter strings. They were instructed that letter order mattered (i.e., ‘a-b’ does not occur in ‘kbac’), while letter contiguity did not (i.e., ‘a-b’ occurs in ‘akcb’). Controlling for crowding and eccentricity, bigrams were recognized faster upon decreasing the letter distance. However, when switching the target letter order (meaning the string should be met by a ‘no’ response), shorter letter distances yielded slower responses and more false positives. Neither relative position-coding models nor absolute position-coding models accommodate both these patterns at once. We discuss how a complete account of our effects may instead combine elements from both model types.

An Integrated Bioinformatics and Functional Approach for miRNA Validation.

MicroRNAs (miRNAs) are small (20-24 nucleotides) non-coding ribo-regulatory molecules with significant roles in regulating target mRNA and long non-coding RNAs at transcriptional and post-transcriptional levels. Rapid advancement in the small RNA sequencing methods with integration of degradome sequencing has accelerated the understanding of miRNA-mediated regulatory hubs in plants and yielded extensive annotation of miRNAs and corresponding targets. However, it is becoming clear that large numbers of such annotations are questionable. Therefore, it is imperative to adopt reliable and strict bioinformatics pipelines for miRNA identification. Furthermore, sensitive methods are needed for validation and functional characterization of miRNA and its target(s). In this chapter, we have provided a comprehensive and streamlined methodology for miRNA identification and its functional validation in plants. This includes a combination of various in silico and experimental methodologies. To identify miRNA compendium from large-scale Next-Generation Sequencing (NGS) small RNA datasets, the miR-PREFeR (miRNA PREdiction From small RNA-Seq data) bioinformatics tool has been described. Also, a homology-based search protocol for finding members of a specific miRNA family has been discussed. The chapter also includes techniques to ascertain miRNA:target pair specificity using in silico target prediction from degradome NGS libraries using CleaveLand pipeline, miRNA:target validation by in planta transient assays, 5' RLM-RACE and expression analysis as well as functional techniques like miRNA overexpression, short tandem target mimic and resistant target approaches. The proposed strategy offers a reliable and sensitive way for miRNA:target identification and validation. Additionally, we strongly promulgate the use of multiple methodologies to validate a miRNA as well as its target.

Protein domain-based prediction of drug/compound-target interactions and experimental validation on LIM kinases.

Predictive approaches such as virtual screening have been used in drug discovery with the objective of reducing developmental time and costs. Current machine learning and network-based approaches have issues related to generalization, usability, or model interpretability, especially due to the complexity of target proteins' structure/function, and bias in system training datasets. Here, we propose a new method "DRUIDom" (DRUg Interacting Domain prediction) to identify bio-interactions between drug candidate compounds and targets by utilizing the domain modularity of proteins, to overcome problems associated with current approaches. DRUIDom is composed of two methodological steps. First, ligands/compounds are statistically mapped to structural domains of their target proteins, with the aim of identifying their interactions. As such, other proteins containing the same mapped domain or domain pair become new candidate targets for the corresponding compounds. Next, a million-scale dataset of small molecule compounds, including those mapped to domains in the previous step, are clustered based on their molecular similarities, and their domain associations are propagated to other compounds within the same clusters. Experimentally verified bioactivity data points, obtained from public databases, are meticulously filtered to construct datasets of active/interacting and inactive/non-interacting drug/compound-target pairs (~2.9M data points), and used as training data for calculating parameters of compound-domain mappings, which led to 27,032 high-confidence associations between 250 domains and 8,165 compounds, and a finalized output of ~5 million new compound-protein interactions. DRUIDom is experimentally validated by syntheses and bioactivity analyses of compounds predicted to target LIM-kinase proteins, which play critical roles in the regulation of cell motility, cell cycle progression, and differentiation through actin filament dynamics. We showed that LIMK-inhibitor-2 and its derivatives significantly block the cancer cell migration through inhibition of LIMK phosphorylation and the downstream protein cofilin. One of the derivative compounds (LIMKi-2d) was identified as a promising candidate due to its action on resistant Mahlavu liver cancer cells. The results demonstrated that DRUIDom can be exploited to identify drug candidate compounds for intended targets and to predict new target proteins based on the defined compound-domain relationships. Datasets, results, and the source code of DRUIDom are fully-available at: https://github.com/cansyl/DRUIDom.

Pairwise effects between lipid GWAS genes modulate lipid plasma levels and cellular uptake

Complex traits are characterized by multiple genes and variants acting simultaneously on a phenotype. However, studying the contribution of individual pairs of genes to complex traits has been challenging since human genetics necessitates very large population sizes, while findings from model systems do not always translate to humans. Here, we combine genetics with combinatorial RNAi (coRNAi) to systematically test for pairwise additive effects (AEs) and genetic interactions (GIs) between 30 lipid genome-wide association studies (GWAS) genes. Gene-based burden tests from 240,970 exomes show that in carriers with truncating mutations in both, APOB and either PCSK9 or LPL (“human double knock-outs”) plasma lipid levels change additively. Genetics and coRNAi identify overlapping AEs for 12 additional gene pairs. Overlapping GIs are observed for TOMM40/APOE with SORT1 and NCAN. Our study identifies distinct gene pairs that modulate plasma and cellular lipid levels primarily via AEs and nominates putative drug target pairs for improved lipid-lowering combination therapies.

Utility of Inter-subject Transfer Learning for Wearable-Sensor-Based Joint Torque Prediction Models.

Generalizability between individuals and groups is often a significant hurdle in model development for human subjects research. In the domain of wearable-sensor-controlled exoskeleton devices, the ability to generalize models across subjects or fine-tune more general models to individual subjects is key to enabling widespread adoption of these technologies. Transfer learning techniques applied to machine learning models afford the ability to apply and investigate the viability and utility such knowledge-transfer scenarios. This paper investigates the utility of single- and multi-subject based parameter transfer on LSTM models trained for "sensor-to-joint torque" prediction tasks, with regards to task performance and computational resources required for network training. We find that parameter transfer between both single- and multi-subject models provide useful knowledge transfer, with varying results across specific "source" and "target" subject pairings. This could be leveraged to lower model training time or computational cost in compute-constrained environments or, with further study to understand causal factors of the observed variance in performance across source and target pairings, to minimize data collection and model retraining requirements to select and personalize a generic model for personalized wearable-sensor-based joint torque prediction technologies.

A new framework for experimental design using Bayesian Evidential Learning: The case of wellhead protection area

Groundwater management practices, such as sustainable drinking water extraction or contamination protection, can have significant socioeconomic impacts. A complete uncertainty analysis should ideally be performed to anticipate all possible outcomes and assess any risk. Uncertainties arise as a result of our limited understanding of the physical processes involved, as well as a scarcity of measurement data, whether directly or indirectly related to the physical parameters of interest. In this paper, we use a small number of tracing experiments (predictor) to predict the wellhead protection area (WHPA, target), the shape and extent of which are influenced by the distribution of hydraulic conductivity (K). Our first goal is to make stochastic predictions of the WHPA within the Bayesian Evidential Learning (BEL) framework, which uses machine learning to find a direct relationship between predictor and target. This relationship is learned using a small number of training models (400) drawn from the prior distribution of K. Forward modelling is used to obtain the 400 pairs of simulated predictors and targets. Newly collected field data can then be used directly to predict the approximate posterior distribution of the corresponding WHPA, obviating the need for the traditional step of data inversion. The number and location of data sources (injection wells) influence the posterior WHPA distribution's uncertainty range. Our second goal is to extend BEL to determine the optimal design of data source locations that minimises the WHPA's posterior uncertainty. Because the BEL model, once trained, allows the computation of the posterior uncertainty corresponding to any new input data, experimental design can be done explicitly, without averaging or approximating. We estimate the WHPA's posterior uncertainty range using the Modified Hausdorff Distance (MHD) and Structural Similarity (SSIM) index metrics. Because the breakthrough curves store information on a large area of the K field surrounding the pumping well, increasing the number of injection wells reduces the derived posterior WHPA uncertainty. Our method can also estimate which injection wells are more informative than others, as demonstrated by a k-fold cross-validation procedure. Overall, the application of BEL to experimental design allows for identifying data sources that maximise the information content of any measurement data while keeping budget constraints and computational costs to a minimum.

Agent-based target evaluation and fire doctrine: an aspect-oriented programming view

Threat evaluation is a vital process for any defense system, and it consists of a series of calculation and evaluation based on the inferred capabilities and intents of the targets that aim to give damage to defended assets. Target evaluation is proceeded in a wargame and the aim is to compare weapon target pairs according to a set of criteria. The target evaluation cycle is repeated anytime a new detection is received and when any change happens in the target currently detected. The whole process consists of a set of tasks that are shared between Command and Control units and the tasks require different responsibilities. Each task is succeeded by a specific behavior that is represented as a reasonably ordered set of actions. The task sharing is organized by taking the C2 architectures into account. In this paper, an agent-based command and control entity, which is in charge of target evaluation and giving engagement decision, is designed and it is situated in an air defense simulation environment. The study aims to propose an agent design in military decision-making domain, bringing analytic methods with the first-order logic together, and combine aspect orientation with agent design. The study also improves dynamic aspect management in agent programming using the relation concept.

Comprehensive analysis of miRNA-mRNA/lncRNA during gonadal development of triploid female rainbow trout (Oncorhynchus mykiss)

Chromosomal ploidy manipulation is one of the means to create excellent germplasm. Triploid fish could provide an ideal sterile model for searching of a underlying mechanism of abnormality in meiosis. The complete understanding of the coding and noncoding RNAs regulating sterility caused by meiosis abnormality is still not well understood. By high-throughput sequencing, we compared the expression profiles of gonadal mRNA, long non-coding RNA (lncRNA), and microRNA (miRNA) at three different developmental stages between the diploid (XX) and triploid (XXX) female rainbow trout. These stages were gonads before differentiation (65 days post fertilisation, dpf), at the beginning of morphological differences (180 dpf) and showing clear difference between diploids and triploids (600 dpf), respectively. A majority of differentially expressed (DE) RNAs were identified, and 22 DE mRNAs related to oocyte meiosis and homologous recombination were characterized. The predicted miRNA-mRNA/lncRNA networks of 3 developmental stages were constructed based on the target pairs of DE lncRNA-miRNA and DE mRNA-miRNA. According to the networks, meiosis-related gene of ccne1 was targeted by dre-miR-15a-5p_R + 1, and 6 targeted DE lncRNAs were identified. Also, qRT-PCR was performed to validate the credibility of the network. Overall, this study explored the potential interplay between coding and noncoding RNAs during the gonadal development of polyploid fish. The mRNA, lncRNA and miRNA screened in this study may be helpful to identify the functional elements regulating fertility of rainbow trout, which may provide reference for character improvement in aquaculture.

Regulatory Network Analysis in Estradiol-Treated Human Endothelial Cells.

Background/Aims: Estrogen has been reported to have beneficial effects on vascular biology through direct actions on endothelium. Together with transcription factors, miRNAs are the major drivers of gene expression and signaling networks. The objective of this study was to identify a comprehensive regulatory network (miRNA–transcription factor–downstream genes) that controls the transcriptomic changes observed in endothelial cells exposed to estradiol. Methods: miRNA/mRNA interactions were assembled using our previous microarray data of human umbilical vein endothelial cells (HUVEC) treated with 17β-estradiol (E2) (1 nmol/L, 24 h). miRNA–mRNA pairings and their associated canonical pathways were determined using Ingenuity Pathway Analysis software. Transcription factors were identified among the miRNA-regulated genes. Transcription factor downstream target genes were predicted by consensus transcription factor binding sites in the promoter region of E2-regulated genes by using JASPAR and TRANSFAC tools in Enrichr software. Results: miRNA–target pairings were filtered by using differentially expressed miRNAs and mRNAs characterized by a regulatory relationship according to miRNA target prediction databases. The analysis identified 588 miRNA–target interactions between 102 miRNAs and 588 targets. Specifically, 63 upregulated miRNAs interacted with 295 downregulated targets, while 39 downregulated miRNAs were paired with 293 upregulated mRNA targets. Functional characterization of miRNA/mRNA association analysis highlighted hypoxia signaling, integrin, ephrin receptor signaling and regulation of actin-based motility by Rho among the canonical pathways regulated by E2 in HUVEC. Transcription factors and downstream genes analysis revealed eight networks, including those mediated by JUN and REPIN1, which are associated with cadherin binding and cell adhesion molecule binding pathways. Conclusion: This study identifies regulatory networks obtained by integrative microarray analysis and provides additional insights into the way estradiol could regulate endothelial function in human endothelial cells.

Integration of miRNAs, Degradome, and Transcriptome Omics Uncovers a Complex Regulatory Network and Provides Insights Into Lipid and Fatty Acid Synthesis During Sesame Seed Development.

Sesame (Sesamum indicum L.) has always been known as a health-promoting oilseed crop because of its nutrient-rich oil. In recent years, studies have focused on lipid and fatty acid (FA) biosynthesis in various plants by high-throughput sequencing. Here, we integrated transcriptomics, small RNAs, and the degradome to establish a comprehensive reserve intensive on key regulatory micro RNA (miRNA)-targeting circuits to better understand the transcriptional and translational regulation of the oil biosynthesis mechanism in sesame seed development. Deep sequencing was performed to differentially express 220 miRNAs, including 65 novel miRNAs, in different developmental periods of seeds. GO and integrated KEGG analysis revealed 32 pairs of miRNA targets with negatively correlated expression profiles, of which 12 miRNA-target pairs were further confirmed by RT-PCR. In addition, a regulatory co-expression network was constructed based on the differentially expressed gene (DEG) profiles. The FAD2, LOC10515945, LOC105161564, and LOC105162196 genes were clustered into groups that regulate the accumulation of unsaturated fatty acid (UFA) biosynthesis. The results provide a unique advanced molecular platform for the study of lipid and FA biosynthesis, and this study may serve as a new theoretical reference to obtain increased levels of UFA from higher-quality sesame seed cultivars and other plants.

Attention Generating Target Images with Labels for Unsupervised Domain Adaptation

Domain adaptation has recently become a popular and successful field of solving the dependence of the deep learning model on the training set. Previous works proposed unsupervised learning methods based on Generative Adversarial Networks (GANs), which solved the mixed target domain labels problem caused by the neglect of class-level matching, by generating target pairs. And these well-designed models have successfully achieved great performance on different datasets. Moreover, due to the recent widespread use of attention mechanism, we intuitively introduce attention mechanism to the previous model. Our new structure is based on GANs containing attention mechanism and well-designed upampling module, which make our model more robust. Demonstrated by extensive experiments, our new model outperforms the previous works on several standard domain adaption datasets.

Abstract 173: Efficient representations of tumor diversity with paired DNA-RNA aberrations

Abstract In this work we develop a framework which allows for a systematic analysis of joint DNA and putative downstream RNA effects in cancer data cohorts. Using the Reactome database, we extract gene pairs that are linked by known mechanistic connections. Such pairs, which we refer to as 'Source Target Pairs' or STPs, consist of a source gene for which we examine aberrant activity in the DNA profile, and a target gene that is affected by said source gene, for which we examine aberrant activity in the RNA profile. Using TCGA data for six different cancer types (breast, colon, kidney, liver, lung and prostate), we use mutation and copy number variation information to compile DNA aberrant activity data. For the same cancer cohorts, we use RNASeq gene expression data to quantify RNA aberrant activity via the previous 'divergence' method we have developed. In the divergence framework, normal samples from the same cancer are used to estimate a normal range of expression for target genes of interest and deviation from the normal range is assumed to indicate aberrant activity which may result from upstream DNA aberrations. Then for a given sample, an STP can be represented as a binary variable, indicating presence or absence of joint DNA-RNA aberrant activity. We utilize integer programming to discover a small set of such STPs for each cancer type such that every sample displays aberrant activity in at least one STP. We refer to these reduced STP configurations as 'minimal coverings' of that cancer. These configurations then allow for the quantification of heterogeneity for that cancer type, as well as for phenotypical groups of interest. This is made possible due to the fact that sample to sample variability can be compared via the entropy of the distribution of the minimal covering, where the small number of STPs in such a configuration makes the computation more tractable. Our results reveal many known putative drivers of cancer, as well as identify some novel genes of interest for further consideration. Comparison of heterogeneity across phenotypes of interest show higher entropy in more pathological phenotypes, indicating increasing heterogeneity with severity of disease. Citation Format: Qian Ke, Wikum Dinalankara, Laurent Younes, Donald Geman, Luigi Marchionni. Efficient representations of tumor diversity with paired DNA-RNA aberrations [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 173.

Predicting Drug-Target Interactions Based on the Ensemble Models of Multiple Feature Pairs.

Backgroud: The prediction of drug–target interactions (DTIs) is of great significance in drug development. It is time-consuming and expensive in traditional experimental methods. Machine learning can reduce the cost of prediction and is limited by the characteristics of imbalanced datasets and problems of essential feature selection. Methods: The prediction method based on the Ensemble model of Multiple Feature Pairs (Ensemble-MFP) is introduced. Firstly, three negative sets are generated according to the Euclidean distance of three feature pairs. Then, the negative samples of the validation set/test set are randomly selected from the union set of the three negative sets in the validation set/test set. At the same time, the ensemble model with weight is optimized and applied to the test set. Results: The area under the receiver operating characteristic curve (area under ROC, AUC) in three out of four sub-datasets in gold standard datasets was more than 94.0% in the prediction of new drugs. The effectiveness of the proposed method is also shown with the comparison of state-of-the-art methods and demonstration of predicted drug–target pairs. Conclusion: The Ensemble-MFP can weigh the existing feature pairs and has a good prediction effect for general prediction on new drugs.

The role of proficiency and pair formation method in language-related episodes: A study of young CLIL learners’ interaction

Little is known about young CLIL (Content and Language Integrated Learning) learners’ attention to formal aspects of the target language when engaged in collaborative task-based interaction. Previous research on language-related episodes (LREs) with other populations indicates that certain variables (e.g. target language proficiency or pair formation method) may play a role in the production of LREs. This study investigates the amount, types and resolution of LREs produced by primary education CLIL learners in a collaborative picture-ordering + story-telling task depending on two variables – L2 English proficiency (grade 5 dyads vs. grade 6 dyads) and pairing method (proficiency-matched dyads vs. student self-selected dyads). Findings indicate that young CLIL learners’ interactive behaviour in L2 English, at least in terms of LRE production, does not differ as a consequence of target language proficiency, whereas pair formation method exerts some influence, self-selected pairs producing and resolving more meaning-based LREs. No differences were found for form-focused LREs.