DNA-methylation variability in normal mucosa: a field cancerization marker in patients with adenomatous polyps.

BACKGROUND: The phenomenon of field cancerization reflects the transition of normal cells into those predisposed to cancer. Assessing the scope and intensity of this process in the colon may support risk prediction and colorectal cancer prevention. METHODS: The Swiss Epigenetic Colorectal Cancer Study (SWEPIC) study, encompassing 1111 participants for DNA methylation analysis and a subset of 84 for RNA sequencing, was employed to detect field cancerization in individuals with adenomatous polyps (AP). Methylation variations were evaluated for their discriminative capability, including in external cohorts, genomic localization, clinical correlations, and associated RNA expression patterns. RESULTS: Normal cecal tissue of individuals harboring an AP in the proximal colon manifested dysregulated DNA methylation compared to tissue from healthy individuals at 558 unique loci. Leveraging these adenoma-related differentially variable and methylated CpGs (aDVMCs), our classifier discerned between healthy and AP-adjacent tissues across SWEPIC datasets (cross-validated area under the receiver operating characteristic curve [ROC AUC] = 0.63-0.81), including within age-stratified cohorts. This discriminative capacity was validated in 3 external sets, differentiating healthy from cancer-adjacent tissue (ROC AUC = 0.82-0.88). Notably, aDVMC dysregulation correlated with polyp multiplicity. More than 50% of aDVMCs were significantly associated with age. These aDVMCs were enriched in active regions of the genome (P < .001), and associated genes exhibited altered expression in AP-adjacent tissues. CONCLUSIONS: Our findings underscore the early onset of field cancerization in the right colon during the neoplastic transformation process. A more extensive validation of aDVMC dysregulation as a stratification tool could pave the way for enhanced surveillance approaches, especially given its linkage to adenoma emergence.

DNA hypermethylation driven by DNMT1 and DNMT3A favors tumor immune escape contributing to the aggressiveness of adrenocortical carcinoma.

BACKGROUND: Adrenocortical carcinoma is rare and aggressive endocrine cancer of the adrenal gland. Within adrenocortical carcinoma, a recently described subtype characterized by a CpG island methylator phenotype (CIMP) has been associated with an especially poor prognosis. However, the drivers of CIMP remain unknown. Furthermore, the functional relation between CIMP and poor clinical outcomes of patients with adrenocortical carcinoma stays elusive. RESULTS: Here, we show that CIMP in adrenocortical carcinoma is linked to the increased expression of DNA methyltransferases DNMT1 and DNMT3A driven by a gain of gene copy number and cell hyperproliferation. Importantly, we demonstrate that CIMP contributes to tumor aggressiveness by favoring tumor immune escape. This effect could be at least partially reversed by treatment with the demethylating agent 5-azacytidine. CONCLUSIONS: In sum, our findings suggest that co-treatment with demethylating agents might enhance the efficacy of immunotherapy and could represent a novel therapeutic approach for patients with high CIMP adrenocortical carcinoma.

scROSHI: robust supervised hierarchical identification of single cells.

Identifying cell types based on expression profiles is a pillar of single cell analysis. Existing machine-learning methods identify predictive features from annotated training data, which are often not available in early-stage studies. This can lead to overfitting and inferior performance when applied to new data. To address these challenges we present scROSHI, which utilizes previously obtained cell type-specific gene lists and does not require training or the existence of annotated data. By respecting the hierarchical nature of cell type relationships and assigning cells consecutively to more specialized identities, excellent prediction performance is achieved. In a benchmark based on publicly available PBMC data sets, scROSHI outperforms competing methods when training data are limited or the diversity between experiments is large.

Reversible transitions between noradrenergic and mesenchymal tumor identities define cell plasticity in neuroblastoma.

Noradrenergic and mesenchymal identities have been characterized in neuroblastoma cell lines according to their epigenetic landscapes and core regulatory circuitries. However, their relationship and relative contribution in patient tumors remain poorly defined. We now document spontaneous and reversible plasticity between the two identities, associated with epigenetic reprogramming, in several neuroblastoma models. Interestingly, xenografts with cells from each identity eventually harbor a noradrenergic phenotype suggesting that the microenvironment provides a powerful pressure towards this phenotype. Accordingly, such a noradrenergic cell identity is systematically observed in single-cell RNA-seq of 18 tumor biopsies and 15 PDX models. Yet, a subpopulation of these noradrenergic tumor cells presents with mesenchymal features that are shared with plasticity models, indicating that the plasticity described in these models has relevance in neuroblastoma patients. This work therefore emphasizes that intrinsic plasticity properties of neuroblastoma cells are dependent upon external cues of the environment to drive cell identity.

Systematic comparison of multi-omics survival models reveals a widespread lack of noise resistance.

As observed in several previous studies, integrating more molecular modalities in multi-omics cancer survival models may not always improve model accuracy. In this study, we compared eight deep learning and four statistical integration techniques for survival prediction on 17 multi-omics datasets, examining model performance in terms of overall accuracy and noise resistance. We found that one deep learning method, mean late fusion, and two statistical methods, PriorityLasso and BlockForest, performed best in terms of both noise resistance and overall discriminative and calibration performance. Nevertheless, all methods struggled to adequately handle noise when too many modalities were added. In summary, we confirmed that current multi-omics survival methods are not sufficiently noise resistant. We recommend relying on only modalities for which there is known predictive value for a particular cancer type until models that have stronger noise-resistance properties are developed.

HDAC1 and PRC2 mediate combinatorial control in SPI1/PU.1-dependent gene repression in murine erythroleukaemia.

Although originally described as transcriptional activator, SPI1/PU.1, a major player in haematopoiesis whose alterations are associated with haematological malignancies, has the ability to repress transcription. Here, we investigated the mechanisms underlying gene repression in the erythroid lineage, in which SPI1 exerts an oncogenic function by blocking differentiation. We show that SPI1 represses genes by binding active enhancers that are located in intergenic or gene body regions. HDAC1 acts as a cooperative mediator of SPI1-induced transcriptional repression by deacetylating SPI1-bound enhancers in a subset of genes, including those involved in erythroid differentiation. Enhancer deacetylation impacts on promoter acetylation, chromatin accessibility and RNA pol II occupancy. In addition to the activities of HDAC1, polycomb repressive complex 2 (PRC2) reinforces gene repression by depositing H3K27me3 at promoter sequences when SPI1 is located at enhancer sequences. Moreover, our study identified a synergistic relationship between PRC2 and HDAC1 complexes in mediating the transcriptional repression activity of SPI1, ultimately inducing synergistic adverse effects on leukaemic cell survival. Our results highlight the importance of the mechanism underlying transcriptional repression in leukemic cells, involving complex functional connections between SPI1 and the epigenetic regulators PRC2 and HDAC1.

Deciphering the etiology and role in oncogenic transformation of the CpG island methylator phenotype: a pan-cancer analysis.

Numerous cancer types have shown to present hypermethylation of CpG islands, also known as a CpG island methylator phenotype (CIMP), often associated with survival variation. Despite extensive research on CIMP, the etiology of this variability remains elusive, possibly due to lack of consistency in defining CIMP. In this work, we utilize a pan-cancer approach to further explore CIMP, focusing on 26 cancer types profiled in the Cancer Genome Atlas (TCGA). We defined CIMP systematically and agnostically, discarding any effects associated with age, gender or tumor purity. We then clustered samples based on their most variable DNA methylation values and analyzed resulting patient groups. Our results confirmed the existence of CIMP in 19 cancers, including gliomas and colorectal cancer. We further showed that CIMP was associated with survival differences in eight cancer types and, in five, represented a prognostic biomarker independent of clinical factors. By analyzing genetic and transcriptomic data, we further uncovered potential drivers of CIMP and classified them in four categories: mutations in genes directly involved in DNA demethylation; mutations in histone methyltransferases; mutations in genes not involved in methylation turnover, such as KRAS and BRAF; and microsatellite instability. Among the 19 CIMP-positive cancers, very few shared potential driver events, and those drivers were only IDH1 and SETD2 mutations. Finally, we found that CIMP was strongly correlated with tumor microenvironment characteristics, such as lymphocyte infiltration. Overall, our results indicate that CIMP does not exhibit a pan-cancer manifestation; rather, general dysregulation of CpG DNA methylation is caused by heterogeneous mechanisms.

Exploring Pathway-Based Group Lasso for Cancer Survival Analysis: A Special Case of Multi-Task Learning.

Motivation: The Cox proportional hazard models are widely used in the study of cancer survival. However, these models often meet challenges such as the large number of features and small sample sizes of cancer data sets. While this issue can be partially solved by applying regularization techniques such as lasso, the models still suffer from unsatisfactory predictive power and low stability. Methods: Here, we investigated two methods to improve survival models. Firstly, we leveraged the biological knowledge that groups of genes act together in pathways and regularized both at the group and gene level using latent group lasso penalty term. Secondly, we designed and applied a multi-task learning penalty that allowed us leveraging the relationship between survival models for different cancers. Results: We observed modest improvements over the simple lasso model with the inclusion of latent group lasso penalty for six of the 16 cancer types tested. The addition of a multi-task penalty, which penalized coefficients in pairs of cancers from diverging too greatly, significantly improved accuracy for a single cancer, lung squamous cell carcinoma, while having minimal effect on other cancer types. Conclusion: While the use of pathway information and multi-tasking shows some promise, these methods do not provide a substantial improvement when compared with standard methods.

CHIPIN: ChIP-seq inter-sample normalization based on signal invariance across transcriptionally constant genes.

BACKGROUND: Multiple studies rely on ChIP-seq experiments to assess the effect of gene modulation and drug treatments on protein binding and chromatin structure. However, most methods commonly used for the normalization of ChIP-seq binding intensity signals across conditions, e.g., the normalization to the same number of reads, either assume a constant signal-to-noise ratio across conditions or base the estimates of correction factors on genomic regions with intrinsically different signals between conditions. Inaccurate normalization of ChIP-seq signal may, in turn, lead to erroneous biological conclusions. RESULTS: We developed a new R package, CHIPIN, that allows normalizing ChIP-seq signals across different conditions/samples when spike-in information is not available, but gene expression data are at hand. Our normalization technique is based on the assumption that, on average, no differences in ChIP-seq signals should be observed in the regulatory regions of genes whose expression levels are constant across samples/conditions. In addition to normalizing ChIP-seq signals, CHIPIN provides as output a number of graphs and calculates statistics allowing the user to assess the efficiency of the normalization and qualify the specificity of the antibody used. In addition to ChIP-seq, CHIPIN can be used without restriction on open chromatin ATAC-seq or DNase hypersensitivity data. We validated the CHIPIN method on several ChIP-seq data sets and documented its superior performance in comparison to several commonly used normalization techniques. CONCLUSIONS: The CHIPIN method provides a new way for ChIP-seq signal normalization across conditions when spike-in experiments are not available. The method is implemented in a user-friendly R package available on GitHub: https://github.com/BoevaLab/CHIPIN.

Chromatin Immunoprecipitation Followed by Next-Generation Sequencing (ChIP-Seq) Analysis in Ewing Sarcoma.

ChIP-seq is the method of choice for profiling protein-DNA interactions, and notably for characterizing the landscape of transcription factor binding and histone modifications. This technique has been widely used to study numerous aspects of tumor biology and led to the development of several promising cancer therapies. In Ewing sarcoma research, ChIP-seq provided important insights into the mechanism of action of the major oncogenic fusion protein EWSR1-FLI1 and related epigenetic and transcriptional changes. In this chapter, we provide a detailed pipeline to analyze ChIP-seq experiments from the preprocessing of raw data to tertiary analysis of detected binding sites. We also advise on best practice to prepare tumor samples prior to sequencing.

Targeted Therapy of TERT-Rearranged Neuroblastoma with BET Bromodomain Inhibitor and Proteasome Inhibitor Combination Therapy.

PURPOSE: TERT gene rearrangement with transcriptional superenhancers leads to TERT overexpression and neuroblastoma. No targeted therapy is available for clinical trials in patients with TERT-rearranged neuroblastoma. EXPERIMENTAL DESIGN: Anticancer agents exerting the best synergistic anticancer effects with BET bromodomain inhibitors were identified by screening an FDA-approved oncology drug library. The synergistic effects of the BET bromodomain inhibitor OTX015 and the proteasome inhibitor carfilzomib were examined by immunoblot and flow cytometry analysis. The anticancer efficacy of OTX015 and carfilzomib combination therapy was investigated in mice xenografted with TERT-rearranged neuroblastoma cell lines or patient-derived xenograft (PDX) tumor cells, and the role of TERT reduction in the anticancer efficacy was examined through rescue experiments in mice. RESULTS: The BET bromodomain protein BRD4 promoted TERT-rearranged neuroblastoma cell proliferation through upregulating TERT expression. Screening of an approved oncology drug library identified the proteasome inhibitor carfilzomib as the agent exerting the best synergistic anticancer effects with BET bromodomain inhibitors including OTX015. OTX015 and carfilzomib synergistically reduced TERT protein expression, induced endoplasmic reticulum stress, and induced TERT-rearranged neuroblastoma cell apoptosis which was blocked by TERT overexpression and endoplasmic reticulum stress antagonists. In mice xenografted with TERT-rearranged neuroblastoma cell lines or PDX tumor cells, OTX015 and carfilzomib synergistically blocked TERT expression, induced tumor cell apoptosis, suppressed tumor progression, and improved mouse survival, which was largely reversed by forced TERT overexpression. CONCLUSIONS: OTX015 and carfilzomib combination therapy is likely to be translated into the first clinical trial of a targeted therapy in patients with TERT-rearranged neuroblastoma.

Transcriptional Programs Define Intratumoral Heterogeneity of Ewing Sarcoma at Single-Cell Resolution.

EWSR1-FLI1, the chimeric oncogene specific for Ewing sarcoma (EwS), induces a cascade of signaling events leading to cell transformation. However, it remains elusive how genetically homogeneous EwS cells can drive the heterogeneity of transcriptional programs. Here, we combine independent component analysis of single-cell RNA sequencing data from diverse cell types and model systems with time-resolved mapping of EWSR1-FLI1 binding sites and of open chromatin regions to characterize dynamic cellular processes associated with EWSR1-FLI1 activity. We thus define an exquisitely specific and direct enhancer-driven EWSR1-FLI1 program. In EwS tumors, cell proliferation and strong oxidative phosphorylation metabolism are associated with a well-defined range of EWSR1-FLI1 activity. In contrast, a subpopulation of cells from below and above the intermediary EWSR1-FLI1 activity is characterized by increased hypoxia. Overall, our study reveals sources of intratumoral heterogeneity within EwS tumors.

CHROMATIX: computing the functional landscape of many-body chromatin interactions in transcriptionally active loci from deconvolved single cells.

Chromatin interactions are important for gene regulation and cellular specialization. Emerging evidence suggests many-body spatial interactions play important roles in condensing super-enhancer regions into a cohesive transcriptional apparatus. Chromosome conformation studies using Hi-C are limited to pairwise, population-averaged interactions; therefore unsuitable for direct assessment of many-body interactions. We describe a computational model, CHROMATIX, which reconstructs ensembles of single-cell chromatin structures by deconvolving Hi-C data and identifies significant many-body interactions. For a diverse set of highly active transcriptional loci with at least 2 super-enhancers, we detail the many-body functional landscape and show DNase accessibility, POLR2A binding, and decreased H3K27me3 are predictive of interaction-enriched regions.

Steroidogenic differentiation and PKA signaling are programmed by histone methyltransferase EZH2 in the adrenal cortex.

Adrenal cortex steroids are essential for body homeostasis, and adrenal insufficiency is a life-threatening condition. Adrenal endocrine activity is maintained through recruitment of subcapsular progenitor cells that follow a unidirectional differentiation path from zona glomerulosa to zona fasciculata (zF). Here, we show that this unidirectionality is ensured by the histone methyltransferase EZH2. Indeed, we demonstrate that EZH2 maintains adrenal steroidogenic cell differentiation by preventing expression of GATA4 and WT1 that cause abnormal dedifferentiation to a progenitor-like state in Ezh2 KO adrenals. EZH2 further ensures normal cortical differentiation by programming cells for optimal response to adrenocorticotrophic hormone (ACTH)/PKA signaling. This is achieved by repression of phosphodiesterases PDE1B, 3A, and 7A and of PRKAR1B. Consequently, EZH2 ablation results in blunted zF differentiation and primary glucocorticoid insufficiency. These data demonstrate an all-encompassing role for EZH2 in programming steroidogenic cells for optimal response to differentiation signals and in maintaining their differentiated state.

TBX2 is a neuroblastoma core regulatory circuitry component enhancing MYCN/FOXM1 reactivation of DREAM targets.

Chromosome 17q gains are almost invariably present in high-risk neuroblastoma cases. Here, we perform an integrative epigenomics search for dosage-sensitive transcription factors on 17q marked by H3K27ac defined super-enhancers and identify TBX2 as top candidate gene. We show that TBX2 is a constituent of the recently established core regulatory circuitry in neuroblastoma with features of a cell identity transcription factor, driving proliferation through activation of p21-DREAM repressed FOXM1 target genes. Combined MYCN/TBX2 knockdown enforces cell growth arrest suggesting that TBX2 enhances MYCN sustained activation of FOXM1 targets. Targeting transcriptional addiction by combined CDK7 and BET bromodomain inhibition shows synergistic effects on cell viability with strong repressive effects on CRC gene expression and p53 pathway response as well as several genes implicated in transcriptional regulation. In conclusion, we provide insight into the role of the TBX2 CRC gene in transcriptional dependency of neuroblastoma cells warranting clinical trials using BET and CDK7 inhibitors.

Meta-mining of copy number profiles of high-risk neuroblastoma tumors.

Neuroblastoma, a pediatric tumor of the sympathetic nervous system, is predominantly driven by copy number aberrations, which predict survival outcome in global neuroblastoma cohorts and in low-risk cases. For high-risk patients there is still a need for better prognostic biomarkers. Via an international collaboration, we collected copy number profiles of 556 high-risk neuroblastomas generated on different array platforms. This manuscript describes the composition of the dataset, the methods used to process the data, including segmentation and aberration calling, and data validation. t-SNE analysis shows that samples cluster according to MYCN status, and shows a difference between array platforms. 97.3% of samples are characterized by the presence of segmental aberrations, in regions frequently affected in neuroblastoma. Focal aberrations affect genes known to be involved in neuroblastoma, such as ALK and LIN28B. To conclude, we compiled a unique large copy number dataset of high-risk neuroblastoma tumors, available via R2 and a Shiny web application. The availability of patient survival data allows to further investigate the prognostic value of copy number aberrations.

FUN-LDA: A Latent Dirichlet Allocation Model for Predicting Tissue-Specific Functional Effects of Noncoding Variation: Methods and Applications.

We describe a method based on a latent Dirichlet allocation model for predicting functional effects of noncoding genetic variants in a cell-type- and/or tissue-specific way (FUN-LDA). Using this unsupervised approach, we predict tissue-specific functional effects for every position in the human genome in 127 different tissues and cell types. We demonstrate the usefulness of our predictions by using several validation experiments. Using eQTL data from several sources, including the GTEx project, Geuvadis project, and TwinsUK cohort, we show that eQTLs in specific tissues tend to be most enriched among the predicted functional variants in relevant tissues in Roadmap. We further show how these integrated functional scores can be used for (1) deriving the most likely cell or tissue type causally implicated for a complex trait by using summary statistics from genome-wide association studies and (2) estimating a tissue-based correlation matrix of various complex traits. We found large enrichment of heritability in functional components of relevant tissues for various complex traits, and FUN-LDA yielded higher enrichment estimates than existing methods. Finally, using experimentally validated functional variants from the literature and variants possibly implicated in disease by previous studies, we rigorously compare FUN-LDA with state-of-the-art functional annotation methods and show that FUN-LDA has better prediction accuracy and higher resolution than these methods. In particular, our results suggest that tissue- and cell-type-specific functional prediction methods tend to have substantially better prediction accuracy than organism-level prediction methods. Scores for each position in the human genome and for each ENCODE and Roadmap tissue are available online (see Web Resources).

Genomic Amplifications and Distal 6q Loss: Novel Markers for Poor Survival in High-risk Neuroblastoma Patients.

Background: Neuroblastoma is characterized by substantial clinical heterogeneity. Despite intensive treatment, the survival rates of high-risk neuroblastoma patients are still disappointingly low. Somatic chromosomal copy number aberrations have been shown to be associated with patient outcome, particularly in low- and intermediate-risk neuroblastoma patients. To improve outcome prediction in high-risk neuroblastoma, we aimed to design a prognostic classification method based on copy number aberrations. Methods: In an international collaboration, normalized high-resolution DNA copy number data (arrayCGH and SNP arrays) from 556 high-risk neuroblastomas obtained at diagnosis were collected from nine collaborative groups and segmented using the same method. We applied logistic and Cox proportional hazard regression to identify genomic aberrations associated with poor outcome. Results: In this study, we identified two types of copy number aberrations that are associated with extremely poor outcome. Distal 6q losses were detected in 5.9% of patients and were associated with a 10-year survival probability of only 3.4% (95% confidence interval [CI] = 0.5% to 23.3%, two-sided P = .002). Amplifications of regions not encompassing the MYCN locus were detected in 18.1% of patients and were associated with a 10-year survival probability of only 5.8% (95% CI = 1.5% to 22.2%, two-sided P < .001). Conclusions: Using a unique large copy number data set of high-risk neuroblastoma cases, we identified a small subset of high-risk neuroblastoma patients with extremely low survival probability that might be eligible for inclusion in clinical trials of new therapeutics. The amplicons may also nominate alternative treatments that target the amplified genes.

Activated ALK signals through the ERK-ETV5-RET pathway to drive neuroblastoma oncogenesis.

Activating mutations of the ALK receptor occur in a subset of neuroblastoma tumors. We previously demonstrated that Alk mutations cooperate with MYCN overexpression to induce neuroblastoma in mice and identified Ret as being strongly upregulated in MYCN/Alkmut tumors. By a genetic approach in vivo, we now document an oncogenic cooperation between activated Ret and MYCN overexpression in neuroblastoma formation. We show that MYCN/RetM919T tumors exhibit histological features and expression profiles close to MYCN/Alkmut tumors. We show that RET transcript levels decrease precedes RET protein levels decrease upon ALK inhibition in neuroblastoma cell lines. Etv5 was identified as a candidate transcription factor regulating Ret expression from murine MYCN/Alkmut tumor transcriptomic data. We demonstrate that ETV5 is regulated both at the protein and mRNA levels upon ALK activation or inhibition in neuroblastoma cell lines and that this regulation precedes RET modulation. We document that ALK activation induces ETV5 protein upregulation through stabilization in a MEK/ERK-dependent manner. We show that RNAi-mediated inhibition of ETV5 decreases RET expression. Reporter assays indicate that ETV5 is able to drive RET gene transcription. ChIP-seq analysis confirmed ETV5 binding on the RET promoter and identified an enhancer upstream of the promoter. Finally, we demonstrate that combining RET and ALK inhibitors reduces tumor growth more efficiently than each single agent in MYCN and AlkF1178L-driven murine neuroblastoma. Altogether, these results define the ERK-ETV5-RET pathway as a critical axis driving neuroblastoma oncogenesis downstream of activated ALK.

QuantumClone: clonal assessment of functional mutations in cancer based on a genotype-aware method for clonal reconstruction.

Motivation: In cancer, clonal evolution is assessed based on information coming from single nucleotide variants and copy number alterations. Nonetheless, existing methods often fail to accurately combine information from both sources to truthfully reconstruct clonal populations in a given tumor sample or in a set of tumor samples coming from the same patient. Moreover, previously published methods detect clones from a single set of variants. As a result, compromises have to be done between stringent variant filtering [reducing dispersion in variant allele frequency estimates (VAFs)] and using all biologically relevant variants. Results: We present a framework for defining cancer clones using most reliable variants of high depth of coverage and assigning functional mutations to the detected clones. The key element of our framework is QuantumClone, a method for variant clustering into clones based on VAFs, genotypes of corresponding regions and information about tumor purity. We validated QuantumClone and our framework on simulated data. We then applied our framework to whole genome sequencing data for 19 neuroblastoma trios each including constitutional, diagnosis and relapse samples. We confirmed an enrichment of damaging variants within such pathways as MAPK (mitogen-activated protein kinases), neuritogenesis, epithelial-mesenchymal transition, cell survival and DNA repair. Most pathways had more damaging variants in the expanding clones compared to shrinking ones, which can be explained by the increased total number of variants between these two populations. Functional mutational rate varied for ancestral clones and clones shrinking or expanding upon treatment, suggesting changes in clone selection mechanisms at different time points of tumor evolution. Availability and implementation: Source code and binaries of the QuantumClone R package are freely available for download at https://CRAN.R-project.org/package=QuantumClone. Contact: gudrun.schleiermacher@curie.fr or valentina.boeva@inserm.fr. Supplementary information: Supplementary data are available at Bioinformatics online.