A comprehensive benchmarking of machine learning algorithms and dimensionality reduction methods for drug sensitivity prediction.

A major challenge of precision oncology is the identification and prioritization of suitable treatment options based on molecular biomarkers of the considered tumor. In pursuit of this goal, large cancer cell line panels have successfully been studied to elucidate the relationship between cellular features and treatment response. Due to the high dimensionality of these datasets, machine learning (ML) is commonly used for their analysis. However, choosing a suitable algorithm and set of input features can be challenging. We performed a comprehensive benchmarking of ML methods and dimension reduction (DR) techniques for predicting drug response metrics. Using the Genomics of Drug Sensitivity in Cancer cell line panel, we trained random forests, neural networks, boosting trees and elastic nets for 179 anti-cancer compounds with feature sets derived from nine DR approaches. We compare the results regarding statistical performance, runtime and interpretability. Additionally, we provide strategies for assessing model performance compared with a simple baseline model and measuring the trade-off between models of different complexity. Lastly, we show that complex ML models benefit from using an optimized DR strategy, and that standard models-even when using considerably fewer features-can still be superior in performance.

miRTargetLink 2.0-interactive miRNA target gene and target pathway networks.

Which genes, gene sets or pathways are regulated by certain miRNAs? Which miRNAs regulate a particular target gene or target pathway in a certain physiological context? Answering such common research questions can be time consuming and labor intensive. Especially for researchers without computational experience, the integration of different data sources, selection of the right parameters and concise visualization can be demanding. A comprehensive analysis should be central to present adequate answers to complex biological questions. With miRTargetLink 2.0, we develop an all-in-one solution for human, mouse and rat miRNA networks. Users input in the unidirectional search mode either a single gene, gene set or gene pathway, alternatively a single miRNA, a set of miRNAs or an miRNA pathway. Moreover, genes and miRNAs can jointly be provided to the tool in the bidirectional search mode. For the selected entities, interaction graphs are generated from different data sources and dynamically presented. Connected application programming interfaces (APIs) to the tailored enrichment tools miEAA and GeneTrail facilitate downstream analysis of pathways and context-annotated categories of network nodes. MiRTargetLink 2.0 is freely accessible at https://www.ccb.uni-saarland.de/mirtargetlink2.

Validation of human microRNA target pathways enables evaluation of target prediction tools.

MicroRNAs are regulators of gene expression. A wide-spread, yet not validated, assumption is that the targetome of miRNAs is non-randomly distributed across the transcriptome and that targets share functional pathways. We developed a computational and experimental strategy termed high-throughput miRNA interaction reporter assay (HiTmIR) to facilitate the validation of target pathways. First, targets and target pathways are predicted and prioritized by computational means to increase the specificity and positive predictive value. Second, the novel webtool miRTaH facilitates guided designs of reporter assay constructs at scale. Third, automated and standardized reporter assays are performed. We evaluated HiTmIR using miR-34a-5p, for which TNF- and TGFB-signaling, and Parkinson's Disease (PD)-related categories were identified and repeated the pipeline for miR-7-5p. HiTmIR validated 58.9% of the target genes for miR-34a-5p and 46.7% for miR-7-5p. We confirmed the targeting by measuring the endogenous protein levels of targets in a neuronal cell model. The standardized positive and negative targets are collected in the new miRATBase database, representing a resource for training, or benchmarking new target predictors. Applied to 88 target predictors with different confidence scores, TargetScan 7.2 and miRanda outperformed other tools. Our experiments demonstrate the efficiency of HiTmIR and provide evidence for an orchestrated miRNA-gene targeting.

The DGCR8 E518K mutation found in Wilms tumors leads to a partial miRNA processing defect that alters gene expression patterns and biological processes.

Wilms tumor (WT) is the most common renal tumor in childhood. We and others have previously identified oncogenic driver mutations affecting the microprocessor genes DROSHA and DGCR8 that lead to altered miRNA expression patterns. In the case of DGCR8, a single recurrent hotspot mutation (E518K) was found in the RNA binding domain. To functionally assess this mutation in vitro, we generated mouse Dgcr8-KO embryonic stem cell (mESC) lines with an inducible expression of wild-type or mutant DGCR8, mirroring the hemizygous mutant expression seen in WT. RNA-seq analysis revealed significant differences of miRNA expression profiles in DGCR8-E518K compared with DGCR8-wild-type mESCs. The E518K mutation only led to a partial rescue of the reported miRNA processing defect in Dgcr8-KO, with selectively reduced expression of numerous canonical miRNAs. Nevertheless, DGCR8-E518K retained significant activity given its ability to still process many miRNAs. Subsequent to altered miRNA levels, the expression of mRNA targets was likewise changed. Functional assays showed that DGCR8-E518K cells still have a partial proliferation and differentiation defect but were able to rescue critical biological processes in embryoid body development. The stem cell program could be shut down and all three germ layers were formed. These findings suggest that the E518K mutation leads to a partial reduction of microprocessor activity and altered specificity with selective impairment only in certain developmental contexts, apparently including nephrogenesis.

MERIDA: a novel Boolean logic-based integer linear program for personalized cancer therapy.

MOTIVATION: A major goal of personalized medicine in oncology is the optimization of treatment strategies given measurements of the genetic and molecular profiles of cancer cells. To further our knowledge on drug sensitivity, machine learning techniques are commonly applied to cancer cell line panels. RESULTS: We present a novel integer linear programming formulation, called MEthod for Rule Identification with multi-omics DAta (MERIDA), for predicting the drug sensitivity of cancer cells. The method represents a modified version of the LOBICO method and yields easily interpretable models amenable to a Boolean logic-based interpretation. Since the proposed altered logical rules lead to an enormous acceleration of the running times of MERIDA compared to LOBICO, we cannot only consider larger input feature sets integrated from genetic and molecular omics data but also build more comprehensive models that mirror the complexity of cancer initiation and progression. Moreover, we enable the inclusion of a priori knowledge that can either stem from biomarker databases or can also be newly acquired knowledge gathered iteratively by previous runs of MERIDA. Our results show that this approach does not only lead to an improved predictive performance but also identifies a variety of putative sensitivity and resistance biomarkers. We also compare our approach to state-of-the-art machine learning methods and demonstrate the superior performance of our method. Hence, MERIDA has great potential to deepen our understanding of the molecular mechanisms causing drug sensitivity or resistance. AVAILABILITY AND IMPLEMENTATION: The corresponding code is available on github (https://github.com/unisb-bioinf/MERIDA.git). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

miRPathDB 2.0: a novel release of the miRNA Pathway Dictionary Database.

Since the initial release of miRPathDB, tremendous progress has been made in the field of microRNA (miRNA) research. New miRNA reference databases have emerged, a vast amount of new miRNA candidates has been discovered and the number of experimentally validated target genes has increased considerably. Hence, the demand for a major upgrade of miRPathDB, including extended analysis functionality and intuitive visualizations of query results has emerged. Here, we present the novel release 2.0 of the miRNA Pathway Dictionary Database (miRPathDB) that is freely accessible at https://mpd.bioinf.uni-sb.de/. miRPathDB 2.0 comes with a ten-fold increase of pre-processed data. In total, the updated database provides putative associations between 27 452 (candidate) miRNAs, 28 352 targets and 16 833 pathways for Homo sapiens, as well as interactions of 1978 miRNAs, 24 898 targets and 6511 functional categories for Mus musculus. Additionally, we analyzed publications citing miRPathDB to identify common use-cases and further extensions. Based on this evaluation, we added new functionality for interactive visualizations and down-stream analyses of bulk queries. In summary, the updated version of miRPathDB, with its new custom-tailored features, is one of the most comprehensive and advanced resources for miRNAs and their target pathways.

Quantitative and time-resolved miRNA pattern of early human T cell activation.

T cells are central to the immune response against various pathogens and cancer cells. Complex networks of transcriptional and post-transcriptional regulators, including microRNAs (miRNAs), coordinate the T cell activation process. Available miRNA datasets, however, do not sufficiently dissolve the dynamic changes of miRNA controlled networks upon T cell activation. Here, we established a quantitative and time-resolved expression pattern for the entire miRNome over a period of 24 h upon human T-cell activation. Based on our time-resolved datasets, we identified central miRNAs and specified common miRNA expression profiles. We found the most prominent quantitative expression changes for miR-155-5p with a range from initially 40 molecules/cell to 1600 molecules/cell upon T-cell activation. We established a comprehensive dynamic regulatory network of both the up- and downstream regulation of miR-155. Upstream, we highlight IRF4 and its complexes with SPI1 and BATF as central for the transcriptional regulation of miR-155. Downstream of miR-155-5p, we verified 17 of its target genes by the time-resolved data recorded after T cell activation. Our data provide comprehensive insights into the range of stimulus induced miRNA abundance changes and lay the ground to identify efficient points of intervention for modifying the T cell response.

GeneTrail 3: advanced high-throughput enrichment analysis.

We present GeneTrail 3, a major extension of our web service GeneTrail that offers rich functionality for the identification, analysis, and visualization of deregulated biological processes. Our web service provides a comprehensive collection of biological processes and signaling pathways for 12 model organisms that can be analyzed with a powerful framework for enrichment and network analysis of transcriptomic, miRNomic, proteomic, and genomic data sets. Moreover, GeneTrail offers novel workflows for the analysis of epigenetic marks, time series experiments, and single cell data. We demonstrate the capabilities of our web service in two case-studies, which highlight that GeneTrail is well equipped for uncovering complex molecular mechanisms. GeneTrail is freely accessible at: http://genetrail.bioinf.uni-sb.de.

The sncRNA Zoo: a repository for circulating small noncoding RNAs in animals.

The repertoire of small noncoding RNAs (sncRNAs), particularly miRNAs, in animals is considered to be evolutionarily conserved. Studies on sncRNAs are often largely based on homology-based information, relying on genomic sequence similarity and excluding actual expression data. To obtain information on sncRNA expression (including miRNAs, snoRNAs, YRNAs and tRNAs), we performed low-input-volume next-generation sequencing of 500 pg of RNA from 21 animals at two German zoological gardens. Notably, none of the species under investigation were previously annotated in any miRNA reference database. Sequencing was performed on blood cells as they are amongst the most accessible, stable and abundant sources of the different sncRNA classes. We evaluated and compared the composition and nature of sncRNAs across the different species by computational approaches. While the distribution of sncRNAs in the different RNA classes varied significantly, general evolutionary patterns were maintained. In particular, miRNA sequences and expression were found to be even more conserved than previously assumed. To make the results available for other researchers, all data, including expression profiles at the species and family levels, and different tools for viewing, filtering and searching the data are freely available in the online resource ASRA (Animal sncRNA Atlas) at https://www.ccb.uni-saarland.de/asra/.

An estimate of the total number of true human miRNAs.

While the number of human miRNA candidates continuously increases, only a few of them are completely characterized and experimentally validated. Toward determining the total number of true miRNAs, we employed a combined in silico high- and experimental low-throughput validation strategy. We collected 28 866 human small RNA sequencing data sets containing 363.7 billion sequencing reads and excluded falsely annotated and low quality data. Our high-throughput analysis identified 65% of 24 127 mature miRNA candidates as likely false-positives. Using northern blotting, we experimentally validated miRBase entries and novel miRNA candidates. By exogenous overexpression of 108 precursors that encode 205 mature miRNAs, we confirmed 68.5% of the miRBase entries with the confirmation rate going up to 94.4% for the high-confidence entries and 18.3% of the novel miRNA candidates. Analyzing endogenous miRNAs, we verified the expression of 8 miRNAs in 12 different human cell lines. In total, we extrapolated 2300 true human mature miRNAs, 1115 of which are currently annotated in miRBase V22. The experimentally validated miRNAs will contribute to revising targetomes hypothesized by utilizing falsely annotated miRNAs.

ClinOmicsTrailbc: a visual analytics tool for breast cancer treatment stratification.

MOTIVATION: Breast cancer is the second leading cause of cancer death among women. Tumors, even of the same histopathological subtype, exhibit a high genotypic diversity that impedes therapy stratification and that hence must be accounted for in the treatment decision-making process. RESULTS: Here, we present ClinOmicsTrailbc, a comprehensive visual analytics tool for breast cancer decision support that provides a holistic assessment of standard-of-care targeted drugs, candidates for drug repositioning and immunotherapeutic approaches. To this end, our tool analyzes and visualizes clinical markers and (epi-)genomics and transcriptomics datasets to identify and evaluate the tumor's main driver mutations, the tumor mutational burden, activity patterns of core cancer-relevant pathways, drug-specific biomarkers, the status of molecular drug targets and pharmacogenomic influences. In order to demonstrate ClinOmicsTrailbc's rich functionality, we present three case studies highlighting various ways in which ClinOmicsTrailbc can support breast cancer precision medicine. ClinOmicsTrailbc is a powerful integrated visual analytics tool for breast cancer research in general and for therapy stratification in particular, assisting oncologists to find the best possible treatment options for their breast cancer patients based on actionable, evidence-based results. AVAILABILITY AND IMPLEMENTATION: ClinOmicsTrailbc can be freely accessed at https://clinomicstrail.bioinf.uni-sb.de. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

miRCarta: a central repository for collecting miRNA candidates.

The continuous increase of available biological data as consequence of modern high-throughput technologies poses new challenges for analysis techniques and database applications. Especially for miRNAs, one class of small non-coding RNAs, many algorithms have been developed to predict new candidates from next-generation sequencing data. While the amount of publications describing novel miRNA candidates keeps steadily increasing, the current gold standard database for miRNAs - miRBase - has not been updated since June 2014. As a result, publications describing new miRNA candidates in the last three to five years might have a substantial overlap of candidates without noticing. With miRCarta we implemented a database to collect novel miRNA candidates and augment the information provided by miRBase. In the first stage, miRCarta is thought to be a highly sensitive collection of potential miRNA candidates with a high degree of analysis functionality, annotations and details on each miRNA. We added-besides the full content of the miRBase-12,857 human miRNA precursors to miRCarta. Users can match their own predictions to the entries of miRCarta to reduce potential redundancies in their studies. miRCarta provides the most comprehensive collection of human miRNAs and miRNA candidates to form a basis for further refinement and validation studies. The database is freely accessible at https://mircarta.cs.uni-saarland.de/.

The role of TCF3 as potential master regulator in blastemal Wilms tumors.

Wilms tumors are the most common type of pediatric kidney tumors. While the overall prognosis for patients is favorable, especially tumors that exhibit a blastemal subtype after preoperative chemotherapy have a poor prognosis. For an improved risk assessment and therapy stratification, it is essential to identify the driving factors that are distinctive for this aggressive subtype. In our study, we compared gene expression profiles of 33 tumor biopsies (17 blastemal and 16 other tumors) after neoadjuvant chemotherapy. The analysis of this dataset using the Regulator Gene Association Enrichment algorithm successfully identified several biomarkers and associated molecular mechanisms that distinguish between blastemal and nonblastemal Wilms tumors. Specifically, regulators involved in embryonic development and epigenetic processes like chromatin remodeling and histone modification play an essential role in blastemal tumors. In this context, we especially identified TCF3 as the central regulatory element. Furthermore, the comparison of ChIP-Seq data of Wilms tumor cell cultures from a blastemal mouse xenograft and a stromal tumor provided further evidence that the chromatin states of blastemal cells share characteristics with embryonic stem cells that are not present in the stromal tumor cell line. These stem-cell like characteristics could potentially add to the increased malignancy and chemoresistance of the blastemal subtype. Along with TCF3, we detected several additional biomarkers that are distinctive for blastemal Wilms tumors after neoadjuvant chemotherapy and that may provide leads for new therapeutic regimens.

A high-resolution map of the human small non-coding transcriptome.

Motivation: Although the amount of small non-coding RNA-sequencing data is continuously increasing, it is still unclear to which extent small RNAs are represented in the human genome. Results: In this study we analyzed 303 billion sequencing reads from nearly 25 000 datasets to answer this question. We determined that 0.8% of the human genome are reliably covered by 874 123 regions with an average length of 31 nt. On the basis of these regions, we found that among the known small non-coding RNA classes, microRNAs were the most prevalent. In subsequent steps, we characterized variations of miRNAs and performed a staged validation of 11 877 candidate miRNAs. Of these, many were actually expressed and significantly dysregulated in lung cancer. Selected candidates were finally validated by northern blots. Although isolated miRNAs could still be present in the human genome, our presented set likely contains the largest fraction of human miRNAs. Contact: c.backes@mx.uni-saarland.de or andreas.keller@ccb.uni-saarland.de. Supplementary information: Supplementary data are available at Bioinformatics online.

REGGAE: a novel approach for the identification of key transcriptional regulators.

Motivation: Transcriptional regulators play a major role in most biological processes. Alterations in their activities are associated with a variety of diseases and in particular with tumor development and progression. Hence, it is important to assess the effects of deregulated regulators on pathological processes. Results: Here, we present REGulator-Gene Association Enrichment (REGGAE), a novel method for the identification of key transcriptional regulators that have a significant effect on the expression of a given set of genes, e.g. genes that are differentially expressed between two sample groups. REGGAE uses a Kolmogorov-Smirnov-like test statistic that implicitly combines associations between regulators and their target genes with an enrichment approach to prioritize the influence of transcriptional regulators. We evaluated our method in two different application scenarios, which demonstrate that REGGAE is well suited for uncovering the influence of transcriptional regulators and is a valuable tool for the elucidation of complex regulatory mechanisms. Availability and implementation: REGGAE is freely available at https://regulatortrail.bioinf.uni-sb.de. Supplementary information: Supplementary data are available at Bioinformatics online.

Large-scale validation of miRNAs by disease association, evolutionary conservation and pathway activity.

The validation of microRNAs (miRNAs) identified by next generation sequencing involves amplification-free and hybridization-based detection of transcripts as criteria for confirming valid miRNAs. Since respective validation is frequently not performed, miRNA repositories likely still contain a substantial fraction of false positive candidates while true miRNAs are not stored in the repositories yet. Especially if downstream analyses are performed with these candidates (e.g. target or pathway prediction), the results may be misleading. In the present study, we evaluated 558 mature miRNAs from miRBase and 1,709 miRNA candidates from next generation sequencing experiments by amplification-free hybridization and investigated their distributions in patients with various disease conditions. Notably, the most significant miRNAs in diseases are often not contained in the miRBase. However, these candidates are evolutionary highly conserved. From the expression patterns, target gene and pathway analyses and evolutionary conservation analyses, we were able to shed light on the complexity of miRNAs in humans. Our data also highlight that a more thorough validation of miRNAs identified by next generation sequencing is required. The results are available in miRCarta ( https://mircarta.cs.uni-saarland.de ).

miRPathDB: a new dictionary on microRNAs and target pathways.

In the last decade, miRNAs and their regulatory mechanisms have been intensively studied and many tools for the analysis of miRNAs and their targets have been developed. We previously presented a dictionary on single miRNAs and their putative target pathways. Since then, the number of miRNAs has tripled and the knowledge on miRNAs and targets has grown substantially. This, along with changes in pathway resources such as KEGG, leads to an improved understanding of miRNAs, their target genes and related pathways. Here, we introduce the miRNA Pathway Dictionary Database (miRPathDB), freely accessible at https://mpd.bioinf.uni-sb.de/ With the database we aim to complement available target pathway web-servers by providing researchers easy access to the information which pathways are regulated by a miRNA, which miRNAs target a pathway and how specific these regulations are. The database contains a large number of miRNAs (2595 human miRNAs), different miRNA target sets (14 773 experimentally validated target genes as well as 19 281 predicted targets genes) and a broad selection of functional biochemical categories (KEGG-, WikiPathways-, BioCarta-, SMPDB-, PID-, Reactome pathways, functional categories from gene ontology (GO), protein families from Pfam and chromosomal locations totaling 12 875 categories). In addition to Homo sapiens, also Mus musculus data are stored and can be compared to human target pathways.

RegulatorTrail: a web service for the identification of key transcriptional regulators.

Transcriptional regulators such as transcription factors and chromatin modifiers play a central role in most biological processes. Alterations in their activities have been observed in many diseases, e.g. cancer. Hence, it is of utmost importance to evaluate and assess the effects of transcriptional regulators on natural and pathogenic processes. Here, we present RegulatorTrail, a web service that provides rich functionality for the identification and prioritization of key transcriptional regulators that have a strong impact on, e.g. pathological processes. RegulatorTrail offers eight methods that use regulator binding information in combination with transcriptomic or epigenomic data to infer the most influential regulators. Our web service not only provides an intuitive web interface, but also a well-documented RESTful API that allows for a straightforward integration into third-party workflows. The presented case studies highlight the capabilities of our web service and demonstrate its potential for the identification of influential regulators: we successfully identified regulators that might explain the increased malignancy in metastatic melanoma compared to primary tumors, as well as important regulators in macrophages. RegulatorTrail is freely accessible at: https://regulatortrail.bioinf.uni-sb.de/.

Phylogenomics with paralogs.

Phylogenomics heavily relies on well-curated sequence data sets that comprise, for each gene, exclusively 1:1 orthologos. Paralogs are treated as a dangerous nuisance that has to be detected and removed. We show here that this severe restriction of the data sets is not necessary. Building upon recent advances in mathematical phylogenetics, we demonstrate that gene duplications convey meaningful phylogenetic information and allow the inference of plausible phylogenetic trees, provided orthologs and paralogs can be distinguished with a degree of certainty. Starting from tree-free estimates of orthology, cograph editing can sufficiently reduce the noise to find correct event-annotated gene trees. The information of gene trees can then directly be translated into constraints on the species trees. Although the resolution is very poor for individual gene families, we show that genome-wide data sets are sufficient to generate fully resolved phylogenetic trees, even in the presence of horizontal gene transfer.

Multi-omics enrichment analysis using the GeneTrail2 web service.

MOTIVATION: Gene set analysis has revolutionized the interpretation of high-throughput transcriptomic data. Nowadays, with comprehensive studies that measure multiple -omics from the same sample, powerful tools for the integrative analysis of multi-omics datasets are required. RESULTS: Here, we present GeneTrail2, a web service allowing the integrated analysis of transcriptomic, miRNomic, genomic and proteomic datasets. It offers multiple statistical tests, a large number of predefined reference sets, as well as a comprehensive collection of biological categories and enables direct comparisons between the computed results. We used GeneTrail2 to explore pathogenic mechanisms of Wilms tumors. We not only succeeded in revealing signaling cascades that may contribute to the malignancy of blastemal subtype tumors but also identified potential biomarkers for nephroblastoma with adverse prognosis. The presented use-case demonstrates that GeneTrail2 is well equipped for the integrative analysis of comprehensive -omics data and may help to shed light on complex pathogenic mechanisms in cancer and other diseases. AVAILABILITY AND IMPLEMENTATION: GeneTrail2 can be freely accessed under https://genetrail2.bioinf.uni-sb.de CONTACT: : dstoeckel@bioinf.uni-sb.de SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.