An analysis pipeline for understanding 6-thioguanine effects on a mouse tumour genome.

Mouse tumour models are extensively used as a pre-clinical research tool in the field of oncology, playing an important role in anticancer drugs discovery. Accordingly, in cancer genomics research, the demand for next-generation sequencing (NGS) is increasing, and consequently, the need for data analysis pipelines is likewise growing. Most NGS data analysis solutions to date do not support mouse data or require highly specific configuration for their use. Here, we present a genome analysis pipeline for mouse tumour NGS data including the whole-genome sequence (WGS) data analysis flow for somatic variant discovery, and the RNA-seq data flow for differential expression, functional analysis and neoantigen prediction. The pipeline is based on standards and best practices and integrates mouse genome references and annotations. In a recent study, the pipeline was applied to demonstrate the efficacy of low dose 6-thioguanine (6TG) treatment on low-mutation melanoma in a pre-clinical mouse model. Here, we further this study and describe in detail the pipeline and the results obtained in terms of tumour mutational burden (TMB) and number of predicted neoantigens, and correlate these with 6TG effects on tumour volume. Our pipeline was expanded to include a neoantigen analysis, resulting in neopeptide prediction and MHC class I antigen presentation evaluation. We observed that the number of predicted neoepitopes were more accurate indicators of tumour immune control than TMB. In conclusion, this study demonstrates the usability of the proposed pipeline, and suggests it could be an essential robust genome analysis platform for future mouse genomic analysis.

Coupled Mass-Spectrometry-Based Lipidomics Machine Learning Approach for Early Detection of Clear Cell Renal Cell Carcinoma.

A discovery-based lipid profiling study of serum samples from a cohort that included patients with clear cell renal cell carcinoma (ccRCC) stages I, II, III, and IV (n = 112) and controls (n = 52) was performed using ultraperformance liquid chromatography coupled to quadrupole-time-of-flight mass spectrometry and machine learning techniques. Multivariate models based on support vector machines and the LASSO variable selection method yielded two discriminant lipid panels for ccRCC detection and early diagnosis. A 16-lipid panel allowed discriminating ccRCC patients from controls with 95.7% accuracy in a training set under cross-validation and 77.1% accuracy in an independent test set. A second model trained to discriminate early (I and II) from late (III and IV) stage ccRCC yielded a panel of 26 compounds that classified stage I patients from an independent test set with 82.1% accuracy. Thirteen species, including cholic acid, undecylenic acid, lauric acid, LPC(16:0/0:0), and PC(18:2/18:2), identified with level 1 exhibited significantly lower levels in samples from ccRCC patients compared to controls. Moreover, 3α-hydroxy-5α-androstan-17-one 3-sulfate, cis-5-dodecenoic acid, arachidonic acid, cis-13-docosenoic acid, PI(16:0/18:1), PC(16:0/18:2), and PC(O-16:0/20:4) contributed to discriminate early from late ccRCC stage patients. The results are auspicious for early ccRCC diagnosis after validation of the panels in larger and different cohorts.

A pan-cancer somatic mutation embedding using autoencoders.

BACKGROUND: Next generation sequencing instruments are providing new opportunities for comprehensive analyses of cancer genomes. The increasing availability of tumor data allows to research the complexity of cancer disease with machine learning methods. The large available repositories of high dimensional tumor samples characterised with germline and somatic mutation data requires advance computational modelling for data interpretation. In this work, we propose to analyze this complex data with neural network learning, a methodology that made impressive advances in image and natural language processing. RESULTS: Here we present a tumor mutation profile analysis pipeline based on an autoencoder model, which is used to discover better representations of lower dimensionality from large somatic mutation data of 40 different tumor types and subtypes. Kernel learning with hierarchical cluster analysis are used to assess the quality of the learned somatic mutation embedding, on which support vector machine models are used to accurately classify tumor subtypes. CONCLUSIONS: The learned latent space maps the original samples in a much lower dimension while keeping the biological signals from the original tumor samples. This pipeline and the resulting embedding allows an easier exploration of the heterogeneity within and across tumor types and to perform an accurate classification of tumor samples in the pan-cancer somatic mutation landscape.

An improved catalogue of putative synaptic genes defined exclusively by temporal transcription profiles through an ensemble machine learning approach.

BACKGROUND: Assembly and function of neuronal synapses require the coordinated expression of a yet undetermined set of genes. Previously, we had trained an ensemble machine learning model to assign a probability of having synaptic function to every protein-coding gene in Drosophila melanogaster. This approach resulted in the publication of a catalogue of 893 genes which we postulated to be very enriched in genes with a still undocumented synaptic function. Since then, the scientific community has experimentally identified 79 new synaptic genes. Here we use these new empirical data to evaluate our original prediction. We also implement a series of changes to the training scheme of our model and using the new data we demonstrate that this improves its predictive power. Finally, we added the new synaptic genes to the training set and trained a new model, obtaining a new, enhanced catalogue of putative synaptic genes. RESULTS: The retrospective analysis demonstrate that our original catalogue was significantly enriched in new synaptic genes. When the changes to the training scheme were implemented using the original training set we obtained even higher enrichment. Finally, applying the new training scheme with a training set including the 79 new synaptic genes, resulted in an enhanced catalogue of putative synaptic genes. Here we present this new catalogue and announce that a regularly updated version will be available online at: http://synapticgenes.bnd.edu.uy CONCLUSIONS: We show that training an ensemble of machine learning classifiers solely with the whole-body temporal transcription profiles of known synaptic genes resulted in a catalogue with a significant enrichment in undiscovered synaptic genes. Using new empirical data provided by the scientific community, we validated our original approach, improved our model an obtained an arguably more precise prediction. This approach reduces the number of genes to be tested through hypothesis-driven experimentation and will facilitate our understanding of neuronal function. AVAILABILITY: http://synapticgenes.bnd.edu.uy.

GenIO: a phenotype-genotype analysis web server for clinical genomics of rare diseases.

BACKGROUND: GenIO is a novel web-server, designed to assist clinical genomics researchers and medical doctors in the diagnostic process of rare genetic diseases. The tool identifies the most probable variants causing a rare disease, using the genomic and clinical information provided by a medical practitioner. Variants identified in a whole-genome, whole-exome or target sequencing studies are annotated, classified and filtered by clinical significance. Candidate genes associated with the patient's symptoms, suspected disease and complementary findings are identified to obtain a small manageable number of the most probable recessive and dominant candidate gene variants associated with the rare disease case. Additionally, following the American College of Medical Genetics and Genomics and the Association of Molecular Pathology (ACMG-AMP) guidelines and recommendations, all potentially pathogenic variants that might be contributing to disease and secondary findings are identified. RESULTS: A retrospective study was performed on 40 patients with a diagnostic rate of 40%. All the known genes that were previously considered as disease causing were correctly identified in the final inherit model output lists. In previously undiagnosed cases, we had no additional yield. CONCLUSION: This unique, intuitive and user-friendly tool to assists medical doctors in the clinical genomics diagnostic process is openly available at https://bioinformatics.ibioba-mpsp-conicet.gov.ar/GenIO/ .

INSECT 2.0: a web-server for genome-wide cis-regulatory modules prediction.

UNLABELLED: INSECT is a user-friendly web server to predict the occurrence of Cis-Regulatory Modules (CRMs), which control gene expression. Here, we present a new release of INSECT which includes several new features, such as whole genome analysis, nucleosome occupancy predictions, and which provides additional links to third-party functional tools that complement user capabilities, CRM analysis and hypothesis construction. Improvements in the core implementation have led to a faster and more efficient tool. In addition, this new release introduces a new interface designed for a more integrative and dynamic user experience. AVAILABILITY AND IMPLEMENTATION: http://bioinformatics.ibioba-mpsp-conicet.gov.ar/INSECT2 CONTACT: pyankilevich@ibioba-mpsp-conicet.gov.ar.

INSECT: IN-silico SEarch for Co-occurring Transcription factors.

MOTIVATION: Transcriptional regulation occurs through the concerted actions of multiple transcription factors (TFs) that bind cooperatively to cis-regulatory modules (CRMs) of genes. These CRMs usually contain a variable number of transcription factor-binding sites (TFBSs) involved in related cellular and physiological processes. Chromatin immunoprecipitation followed by sequencing (ChIP-seq) has been effective in detecting TFBSs and nucleosome location to identify potential CRMs in genome-wide studies. Although several attempts were previously reported to predict the potential binding of TFs at TFBSs within CRMs by comparing different ChIP-seq data, these have been hampered by excessive background, usually emerging as a consequence of experimental conditions. To understand these complex regulatory circuits, it would be helpful to have reliable and updated user-friendly tools to assist in the identification of TFBSs and CRMs for gene(s) of interest. RESULTS: Here we present INSECT (IN-silico SEarch for Co-occurring Transcription factors), a novel web server for identifying potential TFBSs and CRMs in gene sequences. By combining several strategies, INSECT provides flexible analysis of multiple co-occurring TFBSs, by applying differing search schemes and restriction parameters. availability and implementation: INSECT is freely available as a web server at http://bioinformatics.ibioba-mpsp-conicet.gov.ar/INSECT .

Progesterone receptor assembly of a transcriptional complex along with activator protein 1, signal transducer and activator of transcription 3 and ErbB-2 governs breast cancer growth and predicts response to endocrine therapy.

INTRODUCTION: The role of the progesterone receptor (PR) in breast cancer remains a major clinical challenge. Although PR induces mammary tumor growth, its presence in breast tumors is a marker of good prognosis. We investigated coordinated PR rapid and nonclassical transcriptional effects governing breast cancer growth and endocrine therapy resistance. METHODS: We used breast cancer cell lines expressing wild-type and mutant PRs, cells sensitive and resistant to endocrine therapy, a variety of molecular and cellular biology approaches, in vitro proliferation studies and preclinical models to explore PR regulation of cyclin D1 expression, tumor growth, and response to endocrine therapy. We investigated the clinical significance of activator protein 1 (AP-1) and PR interaction in a cohort of 99 PR-positive breast tumors by an immunofluorescence protocol we developed. The prognostic value of AP-1/PR nuclear colocalization in overall survival (OS) was evaluated using Kaplan-Meier method, and Cox model was used to explore said colocalization as an independent prognostic factor for OS. RESULTS: We demonstrated that at the cyclin D1 promoter and through coordinated rapid and transcriptional effects, progestin induces the assembly of a transcriptional complex among AP-1, Stat3, PR, and ErbB-2 which functions as an enhanceosome to drive breast cancer growth. Our studies in a cohort of human breast tumors identified PR and AP-1 nuclear interaction as a marker of good prognosis and better OS in patients treated with tamoxifen (Tam), an anti-estrogen receptor therapy. Rationale for this finding was provided by our demonstration that Tam inhibits rapid and genomic PR effects, rendering breast cancer cells sensitive to its antiproliferative effects. CONCLUSIONS: We here provided novel insight into the paradox of PR action as well as new tools to identify the subgroup of ER+/PR + patients unlikely to respond to ER-targeted therapies.

Thiopurine 6TG treatment increases tumor immunogenicity and response to immune checkpoint blockade.

Immune-checkpoint inhibitors (ICI) are highly effective in reinvigorating T cells to attack cancer. Nevertheless, a large subset of patients fails to benefit from ICI, partly due to lack of the cancer neoepitopes necessary to trigger an immune response. In this study, we used the thiopurine 6-thioguanine (6TG) to induce random mutations and thus increase the level of neoepitopes presented by tumor cells. Thiopurines are prodrugs which are converted into thioguanine nucleotides that are incorporated into DNA (DNA-TG), where they can induce mutation through single nucleotide mismatching. In a pre-clinical mouse model of a mutation-low melanoma cell line, we demonstrated that 6TG induced clinical-grade DNA-TG integration resulting in an improved tumor control that was strongly T cell dependent. 6TG exposure increased the tumor mutational burden, without affecting tumor cell proliferation and cell death. Moreover, 6TG treatment re-shaped the tumor microenvironment by increasing T and NK immune cells, making the tumors more responsive to immune-checkpoint blockade. We further validated that 6TG exposure improved tumor control in additional mouse models of melanoma. These findings have paved the way for a phase I/II clinical trial that explores whether treatment with thiopurines can increase the proportion of otherwise treatment-resistant cancer patients who may benefit from ICI therapy (NCT05276284).

Gene function prediction in five model eukaryotes exclusively based on gene relative location through machine learning.

The function of most genes is unknown. The best results in automated function prediction are obtained with machine learning-based methods that combine multiple data sources, typically sequence derived features, protein structure and interaction data. Even though there is ample evidence showing that a gene's function is not independent of its location, the few available examples of gene function prediction based on gene location rely on sequence identity between genes of different organisms and are thus subjected to the limitations of the relationship between sequence and function. Here we predict thousands of gene functions in five model eukaryotes (Saccharomyces cerevisiae, Caenorhabditis elegans, Drosophila melanogaster, Mus musculus and Homo sapiens) using machine learning models exclusively trained with features derived from the location of genes in the genomes to which they belong. Our aim was not to obtain the best performing method to automated function prediction but to explore the extent to which a gene's location can predict its function in eukaryotes. We found that our models outperform BLAST when predicting terms from Biological Process and Cellular Component Ontologies, showing that, at least in some cases, gene location alone can be more useful than sequence to infer gene function.

Evaluation of T-Cell Responses Against Shared Melanoma Associated Antigens and Predicted Neoantigens in Cutaneous Melanoma Patients Treated With the CSF-470 Allogeneic Cell Vaccine Plus BCG and GM-CSF.

The CSF-470 vaccine consists of lethally-irradiated allogeneic cells derived from four cutaneous melanoma cell lines administered plus BCG and GM-CSF as adjuvants. In an adjuvant phase II study vs. IFN-α2b, the vaccine significantly prolonged the distant metastasis-free survival (DMFS) of stages IIB-IIC-III melanoma patients with evidence of the induction of immune responses against vaccine cells. Purpose: The aim of this study was to analyze the antigens against which the immune response was induced, as well as the T-helper profile and lytic ability of immune cells after CSF-470 treatment. Methods: HLA-restricted peptides from tumor-associated antigens (TAAs) were selected from TANTIGEN database for 13 evaluable vaccinated patients. In addition, for patient #006 (pt#006), tumor somatic variants were identified by NGS and candidate neoAgs were selected by predicted HLA binding affinity and similarity between wild type (wt) and mutant peptides. The patient's PBMC reactivity against selected peptides was detected by IFNγ-ELISPOT. T-helper transcriptional profile was determined by quantifying GATA-3, T-bet, and FOXP3 mRNA by RT-PCR, and intracellular cytokines were analyzed by flow cytometry. Autologous tumor cell lysis by PBMC was assessed in an in vitro calcein release assay. Results: Vaccinated patient's PBMC reactivity against selected TAAs derived peptides showed a progressive increase in the number of IFNγ-producing cells throughout the 2-yr vaccination protocol. ELISPOT response correlated with delayed type hypersensitivity (DTH) reaction to CSF-470 vaccine cells. Early upregulation of GATA-3 and Foxp3 mRNA, as well as an increase in CD4+IL4+cells, was associated with a low DMFS. Also, IFNγ response against 9/73 predicted neoAgs was evidenced in the case of pt#006; 7/9 emerged after vaccination. We verified in pt# 006 that post-vaccination PBMC boosted in vitro with the vaccine lysate were able to lyse autologous tumor cells. Conclusions: A progressive increase in the immune response against TAAs expressed in the vaccine and in the patient's tumor was induced by CSF-470 vaccination. In pt#006, we demonstrated immune recognition of patient's specific neoAgs, which emerged after vaccination. These results suggest that an initial response against shared TAAs could further stimulate an immune response against autologous tumor neoAgs.

Global site-specific neddylation profiling reveals that NEDDylated cofilin regulates actin dynamics.

Neddylation is the post-translational protein modification most closely related to ubiquitination. Whereas the ubiquitin-like protein NEDD8 is well studied for its role in activating cullin-RING E3 ubiquitin ligases, little is known about other substrates. We developed serial NEDD8-ubiquitin substrate profiling (sNUSP), a method that employs NEDD8 R74K knock-in HEK293 cells, allowing discrimination of endogenous NEDD8- and ubiquitin-modification sites by MS after Lys-C digestion and K-εGG-peptide enrichment. Using sNUSP, we identified 607 neddylation sites dynamically regulated by the neddylation inhibitor MLN4924 and the de-neddylating enzyme NEDP1, implying that many non-cullin proteins are neddylated. Among the candidates, we characterized lysine 112 of the actin regulator cofilin as a novel neddylation event. Global inhibition of neddylation in developing neurons leads to cytoskeletal defects, altered actin dynamics and neurite growth impairments, whereas site-specific neddylation of cofilin at K112 regulates neurite outgrowth, suggesting that cofilin neddylation contributes to the regulation of neuronal actin organization.

Publisher Correction: Global site-specific neddylation profiling reveals that NEDDylated cofilin regulates actin dynamics.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

miRNome profiling of LSC-enriched CD34+CD38-CD26+ fraction in Ph+ CML-CP samples from Argentinean patients: a potential new pharmacogenomic tool.

Chronic myeloid leukemia (CML) is a myeloid stem cell neoplasm characterized by an expansion of myeloid progenitor cells and the presence of BCR-ABL1 oncoprotein. Since the introduction of specific BCR-ABL1 tyrosine kinase inhibitors (TKI), overall survival has improved significantly. However, under long-term therapy patients may have residual disease that originates from TKI-resistant leukemic stem cells (LSC). In this work, we analyzed the miRNome of LSC-enriched CD34+CD38-CD26+ and normal hematopoietic stem cells (HSC) fractions obtained from the same chronic phase (CP) CML patients, and stem and progenitor cells obtained from healthy donors (HD) by next-generation sequencing. We detected a global decrease of microRNA levels in LSC-enriched CD34+CD38-CD26+ and HSC fractions from CML-CP patients, and decreased levels of microRNAs and snoRNAs from a genomic cluster in chromosome 14, suggesting a mechanism of silencing of multiple non-coding RNAs. Surprisingly, HSC from CML-CP patients, despite the absence of BCR-ABL1 expression, showed an altered miRNome. We confirmed by RT-qPCR that the levels of miR-196a-5p were increased more than nine-fold in CD26+ (BCR-ABL1 + ) vs. CD26- (BCR-ABL1 -) CD34+CD38- fractions from CML-CP patients at diagnosis, and in silico analysis revealed a significant association to lipid metabolism and hematopoiesis functions. In the light of recent descriptions of increased oxidative metabolism in CML LSC-enriched fractions, these results serve as a guide for future functional studies that evaluate the role of microRNAs in this process. Metabolic vulnerabilities in LSCs open the road for new therapeutic strategies. This is the first report of the miRNome of CML-CP CD34+CD38- fractions that distinguishes between CD26+ (BCR-ABL1 + ) and their CD26- (BCR-ABL1 - ) counterparts, providing valuable data for future studies.

von Hippel-Lindau mutants in renal cell carcinoma are regulated by increased expression of RSUME.

Renal cell carcinoma (RCC) is the major cause of death among patients with von Hippel-Lindau (VHL) disease. Resistance to therapies targeting tumor angiogenesis opens the question about the underlying mechanisms. Previously we have described that RWDD3 or RSUME (RWD domain-containing protein SUMO Enhancer) sumoylates and binds VHL protein and negatively regulates HIF degradation, leading to xenograft RCC tumor growth in mice. In this study, we performed a bioinformatics analysis in a ccRCC dataset showing an association of RSUME levels with VHL mutations and tumor progression, and we demonstrate the molecular mechanism by which RSUME regulates the pathologic angiogenic phenotype of VHL missense mutations. We report that VHL mutants fail to downregulate RSUME protein levels accounting for the increased RSUME expression found in RCC tumors. Furthermore, we prove that targeting RSUME in RCC cell line clones carrying missense VHL mutants results in decreased early tumor angiogenesis. The mechanism we describe is that RSUME sumoylates VHL mutants and beyond its sumoylation capacity, interacts with Type 2 VHL mutants, reduces HIF-2α-VHL mutants binding, and negatively regulates the assembly of the Type 2 VHL, Elongins and Cullins (ECV) complex. Altogether these results show RSUME involvement in VHL mutants deregulation that leads to the angiogenic phenotype of RCC tumors.

Changes in the TCRß Repertoire and Tumor Immune Signature From a Cutaneous Melanoma Patient Immunized With the CSF-470 Vaccine: A Case Report.

The allogeneic therapeutic vaccine CSF-470 has demonstrated a significant benefit over medium-dose IFNα2b in the distant metastasis-free survival for stages IIB-IIC-III cutaneous melanoma patients in a randomized phase II/III clinical trial (CASVAC-0401, NCT01729663). At the end of the 2-year CSF-470 immunization protocol, patient #006 developed several lung and one subcutaneous melanoma metastases; this later was excised. In this report, we analyzed the changes throughout vaccination of immune populations in blood and in the tumor tissue, with special focus on the T-cell repertoire. Immunohistochemistry revealed a marked increase in CD8+, CD4+, and CD20+ lymphocytes infiltrating the metastasis relative to the primary tumor. Lymphocytes were firmly attached to dying-tumor cells containing Granzyme-B granules. Whole-exon sequencing assessment indicated a moderate-to-high tumor mutational burden, with BRAFV600E as the main oncogenic driver. Mutational signature presented large numbers of mutations at dipyrimidines, typical of melanoma. Relevant tumor and immune-related genes from the subcutaneous metastasis were addressed by RNA-Seq analysis, revealing expression of typical melanoma antigens and proliferative tumor-related genes. Stimulatory and inhibitory immune transcripts were detected as well as evidence of active T-cell effector function. Peripheral blood monitoring revealed an increase in CD4+ and CD8+ cells by the end of the immunization protocol. By CDR3-T-cell receptor ß (TCRß) sequencing, generation of new clones and an increase in oligoclonality was observed in the peripheral T-cells immune repertoire throughout immunization. A shift, with the expansion of selected preexisting and newly arising clones with reduction of others, was detected in blood. In tumor-infiltrating lymphocytes, prevalent clones (50%) were both new and preexisting that were expanded in blood following CSF-470 immunization. These clones persisted in time, since 2 years after completing the immunization, 51% of the clones present in the metastasis were still detected in blood. This is the first report of the modulation of the TCRß repertoire from a melanoma patient immunized with the CSF-470 vaccine. After immunization, the changes observed in peripheral immune populations as well as in the tumor compartment suggest that the vaccine can induce an antitumor adaptive immune repertoire that can reach tumor lesions and persists in blood for at least 2 years.

Evaluation of the primitive fraction by functional in vitro assays at the RNA and DNA level represents a novel tool for complementing molecular monitoring in chronic myeloid leukemia.

Quantification of BCR-ABL1 mRNA levels in peripheral blood of chronic myeloid leukemia patients is a strong indicator of response to tyrosine-kinase inhibitors (TKI) treatment. However, additional prognostic markers are needed in order to better classify patients. The hypothesis of leukemic stem cells (LSCs) heterogeneity and persistence, suggests that their functional evaluation could be of clinical interest. In this work, we assessed the primitive and progenitor fractions in patients at diagnosis and during TKI treatment using functional in vitro assays, defining a "functional leukemic burden" (FLB). We observed that the FLB was reduced in vivo in both fractions upon treatment. However, different FLB levels were observed among patients according to their response to treatment, suggesting that quantification of the FLB could complement early molecular monitoring. Given that FLB assessment is limited by BCR-ABL1 mRNA expression levels, we developed a novel detection method of primitive cells at the DNA level, using patient-specific primers and direct nested PCR in colonies obtained from functional in vitro assays. We believe that this method could be useful in the context of discontinuation trials, given that it is unknown whether the persistent leukemic clone represents LSCs, able to resume the leukemia upon TKI removal.

IDconverter and IDClight: conversion and annotation of gene and protein IDs.

BACKGROUND: Researchers involved in the annotation of large numbers of gene, clone or protein identifiers are usually required to perform a one-by-one conversion for each identifier. When the field of research is one such as microarray experiments, this number may be around 30,000. RESULTS: To help researchers map accession numbers and identifiers among clones, genes, proteins and chromosomal positions, we have designed and developed IDconverter and IDClight. They are two user-friendly, freely available web server applications that also provide additional functional information by mapping the identifiers on to pathways, Gene Ontology terms, and literature references. Both tools are high-throughput oriented and include identifiers for the most common genomic databases. These tools have been compared to other similar tools, showing that they are among the fastest and the most up-to-date. CONCLUSION: These tools provide a fast and intuitive way of enriching the information coming out of high-throughput experiments like microarrays. They can be valuable both to wet-lab researchers and to bioinformaticians.

GEPAS, an experiment-oriented pipeline for the analysis of microarray gene expression data.

The Gene Expression Profile Analysis Suite, GEPAS, has been running for more than three years. With >76,000 experiments analysed during the last year and a daily average of almost 300 analyses, GEPAS can be considered a well-established and widely used platform for gene expression microarray data analysis. GEPAS is oriented to the analysis of whole series of experiments. Its design and development have been driven by the demands of the biomedical community, probably the most active collective in the field of microarray users. Although clustering methods have obviously been implemented in GEPAS, our interest has focused more on methods for finding genes differentially expressed among distinct classes of experiments or correlated to diverse clinical outcomes, as well as on building predictors. There is also a great interest in CGH-arrays which fostered the development of the corresponding tool in GEPAS: InSilicoCGH. Much effort has been invested in GEPAS for developing and implementing efficient methods for functional annotation of experiments in the proper statistical framework. Thus, the popular FatiGO has expanded to a suite of programs for functional annotation of experiments, including information on transcription factor binding sites, chromosomal location and tissues. The web-based pipeline for microarray gene expression data, GEPAS, is available at http://www.gepas.org.