Binary classifiers are routinely evaluated with performance measures such as sensitivity and specificity, and performance is frequently illustrated with Receiver Operating Characteristics (ROC) ...plots. Alternative measures such as positive predictive value (PPV) and the associated Precision/Recall (PRC) plots are used less frequently. Many bioinformatics studies develop and evaluate classifiers that are to be applied to strongly imbalanced datasets in which the number of negatives outweighs the number of positives significantly. While ROC plots are visually appealing and provide an overview of a classifier's performance across a wide range of specificities, one can ask whether ROC plots could be misleading when applied in imbalanced classification scenarios. We show here that the visual interpretability of ROC plots in the context of imbalanced datasets can be deceptive with respect to conclusions about the reliability of classification performance, owing to an intuitive but wrong interpretation of specificity. PRC plots, on the other hand, can provide the viewer with an accurate prediction of future classification performance due to the fact that they evaluate the fraction of true positives among positive predictions. Our findings have potential implications for the interpretation of a large number of studies that use ROC plots on imbalanced datasets.
Gene expression is regulated through cis-regulatory elements (CREs), among which are promoters, enhancers, Polycomb/Trithorax Response Elements (PREs), silencers and insulators. Computational ...prediction of CREs can be achieved using a variety of statistical and machine learning methods combined with different feature space formulations. Although Python packages for DNA sequence feature sets and for machine learning are available, no existing package facilitates the combination of DNA sequence feature sets with machine learning methods for the genome-wide prediction of candidate CREs. We here present Gnocis, a Python package that streamlines the analysis and the modelling of CRE sequences by providing extensible APIs and implementing the glue required for combining feature sets and models for genome-wide prediction. Gnocis implements a variety of base feature sets, including motif pair occurrence frequencies and the k-spectrum mismatch kernel. It integrates with Scikit-learn and TensorFlow for state-of-the-art machine learning. Gnocis additionally implements a broad suite of tools for the handling and preparation of sequence, region and curve data, which can be useful for general DNA bioinformatics in Python. We also present Deep-MOCCA, a neural network architecture inspired by SVM-MOCCA that achieves moderate to high generalization without prior motif knowledge. To demonstrate the use of Gnocis, we applied multiple machine learning methods to the modelling of D. melanogaster PREs, including a Convolutional Neural Network (CNN), making this the first study to model PREs with CNNs. The models are readily adapted to new CRE modelling problems and to other organisms. In order to produce a high-performance, compiled package for Python 3, we implemented Gnocis in Cython. Gnocis can be installed using the PyPI package manager by running 'pip install gnocis'. The source code is available on GitHub, at
Gene expression is regulated through cis-regulatory elements (CREs), among which are promoters, enhancers, Polycomb/Trithorax Response Elements (PREs), silencers and insulators. Computational ...prediction of CREs can be achieved using a variety of statistical and machine learning methods combined with different feature space formulations. Although Python packages for DNA sequence feature sets and for machine learning are available, no existing package facilitates the combination of DNA sequence feature sets with machine learning methods for the genome-wide prediction of candidate CREs. We here present Gnocis, a Python package that streamlines the analysis and the modelling of CRE sequences by providing extensible APIs and implementing the glue required for combining feature sets and models for genome-wide prediction. Gnocis implements a variety of base feature sets, including motif pair occurrence frequencies and the k-spectrum mismatch kernel. It integrates with Scikit-learn and TensorFlow for state-of-the-art machine learning. Gnocis additionally implements a broad suite of tools for the handling and preparation of sequence, region and curve data, which can be useful for general DNA bioinformatics in Python. We also present Deep-MOCCA, a neural network architecture inspired by SVM-MOCCA that achieves moderate to high generalization without prior motif knowledge. To demonstrate the use of Gnocis, we applied multiple machine learning methods to the modelling of D. melanogaster PREs, including a Convolutional Neural Network (CNN), making this the first study to model PREs with CNNs. The models are readily adapted to new CRE modelling problems and to other organisms. In order to produce a high-performance, compiled package for Python 3, we implemented Gnocis in Cython. Gnocis can be installed using the PyPI package manager by running 'pip install gnocis'. The source code is available on GitHub, at https://github.com/bjornbredesen/gnocis.
MicroRNAs (miRNAs) are short RNAs that post-transcriptionally regulate the expression of target genes by binding to the target mRNAs. Although a large number of animal miRNAs has been defined, only a ...few targets are known. In contrast to plant miRNAs, which usually bind nearly perfectly to their targets, animal miRNAs bind less tightly, with a few nucleotides being unbound, thus producing more complex secondary structures of miRNA/target duplexes. Here, we present a program, RNA-hybrid, that predicts multiple potential binding sites of miRNAs in large target RNAs. In general, the program finds the energetically most favorable hybridization sites of a small RNA in a large RNA. Intramolecular hybridizations, that is, base pairings between target nucleotides or between miRNA nucleotides are not allowed. For large targets, the time complexity of the algorithm is linear in the target length, allowing many long targets to be searched in a short time. Statistical significance of predicted targets is assessed with an extreme value statistics of length normalized minimum free energies, a Poisson approximation of multiple binding sites, and the calculation of effective numbers of orthologous targets in comparative studies of multiple organisms. We applied our method to the prediction of Drosophila miRNA targets in 3'UTRs and coding sequence. RNAhybrid, with its accompanying programs RNAcalibrate and RNAeffective, is available for download and as a Web tool on the Bielefeld Bioinformatics Server (http://bibiserv.techfak.uni-bielefeld.de/rnahybrid/).
Cis-regulatory elements (CREs) are DNA sequence segments that regulate gene expression. Among CREs are promoters, enhancers, Boundary Elements (BEs) and Polycomb Response Elements (PREs), all of ...which are enriched in specific sequence motifs that form particular occurrence landscapes. We have recently introduced a hierarchical machine learning approach (SVM-MOCCA) in which Support Vector Machines (SVMs) are applied on the level of individual motif occurrences, modelling local sequence composition, and then combined for the prediction of whole regulatory elements. We used SVM-MOCCA to predict PREs in Drosophila and found that it was superior to other methods. However, we did not publish a polished implementation of SVM-MOCCA, which can be useful for other researchers, and we only tested SVM-MOCCA with IUPAC motifs and PREs.
We here present an expanded suite for modelling CRE sequences in terms of motif occurrence combinatorics-Motif Occurrence Combinatorics Classification Algorithms (MOCCA). MOCCA contains efficient implementations of several modelling methods, including SVM-MOCCA, and a new method, RF-MOCCA, a Random Forest-derivative of SVM-MOCCA. We used SVM-MOCCA and RF-MOCCA to model Drosophila PREs and BEs in cross-validation experiments, making this the first study to model PREs with Random Forests and the first study that applies the hierarchical MOCCA approach to the prediction of BEs. Both models significantly improve generalization to PREs and boundary elements beyond that of previous methods-including 4-spectrum and motif occurrence frequency Support Vector Machines and Random Forests-, with RF-MOCCA yielding the best results.
MOCCA is a flexible and powerful suite of tools for the motif-based modelling of CRE sequences in terms of motif composition. MOCCA can be applied to any new CRE modelling problems where motifs have been identified. MOCCA supports IUPAC and Position Weight Matrix (PWM) motifs. For ease of use, MOCCA implements generation of negative training data, and additionally a mode that requires only that the user specifies positives, motifs and a genome. MOCCA is licensed under the MIT license and is available on Github at https://github.com/bjornbredesen/MOCCA .
Total-RNA sequencing (total-RNA-seq) allows the simultaneous study of both the coding and the non-coding transcriptome. Yet, computational pipelines have traditionally focused on particular biotypes, ...making assumptions that are not fullfilled by total-RNA-seq datasets. Transcripts from distinct RNA biotypes vary in length, biogenesis, and function, can overlap in a genomic region, and may be present in the genome with a high copy number. Consequently, reads from total-RNA-seq libraries may cause ambiguous genomic alignments, demanding for flexible quantification approaches.
Here we present Multi-Graph count (MGcount), a total-RNA-seq quantification tool combining two strategies for handling ambiguous alignments. First, MGcount assigns reads hierarchically to small-RNA and long-RNA features to account for length disparity when transcripts overlap in the same genomic position. Next, MGcount aggregates RNA products with similar sequences where reads systematically multi-map using a graph-based approach. MGcount outputs a transcriptomic count matrix compatible with RNA-sequencing downstream analysis pipelines, with both bulk and single-cell resolution, and the graphs that model repeated transcript structures for different biotypes. The software can be used as a python module or as a single-file executable program.
MGcount is a flexible total-RNA-seq quantification tool that successfully integrates reads that align to multiple genomic locations or that overlap with multiple gene features. Its approach is suitable for the simultaneous estimation of protein-coding, long non-coding and small non-coding transcript concentration, in both precursor and processed forms. Both source code and compiled software are available at https://github.com/hitaandrea/MGcount .
The precision-recall plot is more informative than the ROC plot when evaluating classifiers on imbalanced datasets, but fast and accurate curve calculation tools for precision-recall plots are ...currently not available. We have developed Precrec, an R library that aims to overcome this limitation of the plot. Our tool provides fast and accurate precision-recall calculations together with multiple functionalities that work efficiently under different conditions.
Precrec is licensed under GPL-3 and freely available from CRAN (https://cran.r-project.org/package=precrec). It is implemented in R with C ++.
takaya.saito@ii.uib.noSupplementary information: Supplementary data are available at Bioinformatics online.
Response to Wang and Luo Rehmsmeier, Marc
BMC biology,
04/2012, Letnik:
10, Številka:
1
Journal Article
Recenzirano
Odprti dostop
This article is a response to Wang and Luo.See correspondence article http://www.biomedcentral.com/1741-7007/10/30 and the original research article http://www.biomedcentral.com/1741-7007/9/24.
Genomic imprinting is an epigenetic phenomenon leading to parent-of-origin specific differential expression of maternally and paternally inherited alleles. In plants, genomic imprinting has mainly ...been observed in the endosperm, an ephemeral triploid tissue derived after fertilization of the diploid central cell with a haploid sperm cell. In an effort to identify novel imprinted genes in Arabidopsis thaliana, we generated deep sequencing RNA profiles of F1 hybrid seeds derived after reciprocal crosses of Arabidopsis Col-0 and Bur-0 accessions. Using polymorphic sites to quantify allele-specific expression levels, we could identify more than 60 genes with potential parent-of-origin specific expression. By analyzing the distribution of DNA methylation and epigenetic marks established by Polycomb group (PcG) proteins using publicly available datasets, we suggest that for maternally expressed genes (MEGs) repression of the paternally inherited alleles largely depends on DNA methylation or PcG-mediated repression, whereas repression of the maternal alleles of paternally expressed genes (PEGs) predominantly depends on PcG proteins. While maternal alleles of MEGs are also targeted by PcG proteins, such targeting does not cause complete repression. Candidate MEGs and PEGs are enriched for cis-proximal transposons, suggesting that transposons might be a driving force for the evolution of imprinted genes in Arabidopsis. In addition, we find that MEGs and PEGs are significantly faster evolving when compared to other genes in the genome. In contrast to the predominant location of mammalian imprinted genes in clusters, cluster formation was only detected for few MEGs and PEGs, suggesting that clustering is not a major requirement for imprinted gene regulation in Arabidopsis.