DeepCLIP: predicting the effect of mutations on protein-RNA binding with deep learning.

Nucleotide variants can cause functional changes by altering protein-RNA binding in various ways that are not easy to predict. This can affect processes such as splicing, nuclear shuttling, and stability of the transcript. Therefore, correct modeling of protein-RNA binding is critical when predicting the effects of sequence variations. Many RNA-binding proteins recognize a diverse set of motifs and binding is typically also dependent on the genomic context, making this task particularly challenging. Here, we present DeepCLIP, the first method for context-aware modeling and predicting protein binding to RNA nucleic acids using exclusively sequence data as input. We show that DeepCLIP outperforms existing methods for modeling RNA-protein binding. Importantly, we demonstrate that DeepCLIP predictions correlate with the functional outcomes of nucleotide variants in independent wet lab experiments. Furthermore, we show how DeepCLIP binding profiles can be used in the design of therapeutically relevant antisense oligonucleotides, and to uncover possible position-dependent regulation in a tissue-specific manner. DeepCLIP is freely available as a stand-alone application and as a webtool at http://deepclip.compbio.sdu.dk.

Genomic profiling in ovarian cancer retreated with platinum based chemotherapy presented homologous recombination deficiency and copy number imbalances of CCNE1 and RB1 genes.

BACKGROUND: Ovarian carcinomas presenting homologous recombination deficiency (HRD), which is observed in about 50% of cases, are more sensitive to platinum and PARP inhibitor therapies. Although platinum resistant disease has a low chance to be responsive to platinum-based chemotherapy, a set of patients is retreated with platinum and some of them are responsive. In this study, we evaluated copy number alterations, HR gene mutations and HR deficiency scores in ovarian cancer patients with prolonged platinum sensitivity. METHODS: In this retrospective study (2005 to 2014), we selected 31 patients with platinum resistant ovarian cancer retreated with platinum therapy. Copy number alterations and HR scores were evaluated using the OncoScan® FFPE platform in 15 cases. The mutational profile of 24 genes was investigated by targeted-NGS. RESULTS: The median values of the four HRD scores were higher in responders (LOH = 15, LST = 28, tAI = 33, CS = 84) compared with non-responders (LOH = 7.5, LST = 17.5, tAI = 23, CS = 47). Patients with high LOH, LST, tAI and CS scores had better response rates, although these differences were not statistically significant. Response rate to platinum retreatment was 22% in patients with CCNE1 gains and 83.5% in patients with no CCNE1 gains (p = 0.041). Furthermore, response rate was 54.5% in patients with RB1 loss and 25% in patients without RB1 loss (p = 0.569). Patients with CCNE1 gains showed a worse progression free survival (PFS = 11.1 months vs 3.7 months; p = 0.008) and a shorter overall survival (OS = 39.3 months vs 7.1 months; p = 0.007) in comparison with patients with no CCNE1 gains. Patients with RB1 loss had better PFS (9.0 months vs 2.6 months; p = 0.093) and OS (27.4 months vs 3.6 months; p = 0.025) compared with cases with no RB1 loss. Four tumor samples were BRCA mutated and tumor mutations were not associated with response to treatment. CONCLUSIONS: HR deficiency was found in 60% of our cases and HRD medium values were higher in responders than in non-responders. Despite the small number of patients tested, CCNE1 gain and RB1 loss discriminate patients with tumors extremely sensitive to platinum retreatment.

Comprehensive Genomic Profiling of Androgen-Receptor-Negative Canine Prostate Cancer.

Canine carcinomas have been considered natural models for human diseases; however, the genomic profile of canine prostate cancers (PCs) has not been explored. In this study, 14 PC androgen-receptor-negative cases, 4 proliferative inflammatory atrophies (PIA), and 5 normal prostate tissues were investigated by array-based comparative genomic hybridization (aCGH). Copy number alterations (CNAs) were assessed using the Canine Genome CGH Microarray 4 × 44K (Agilent Technologies). Genes covered by recurrent CNAs were submitted to enrichment and cross-validation analysis. In addition, the expression levels of TP53, MDM2 and ZBTB4 were evaluated in an independent set of cases by qPCR. PC cases presented genomic complexity, while PIA samples had a small number of CNAs. Recurrent losses covering well-known tumor suppressor genes, such as ATM, BRCA1, CDH1, MEN1 and TP53, were found in PC. The in silico functional analysis showed several cancer-related genes associated with canonical pathways and interaction networks previously described in human PC. The MDM2, TP53, and ZBTB4 copy number alterations were translated into altered expression levels. A cross-validation analysis using The Cancer Genome Atlas (TCGA) database for human PC uncovered similarities between canine and human PCs. Androgen-receptor-negative canine PC is a complex disease characterized by high genomic instability, showing a set of genes with similar alterations to human cancer.

CoNVaQ: a web tool for copy number variation-based association studies.

BACKGROUND: Copy number variations (CNVs) are large segments of the genome that are duplicated or deleted. Structural variations in the genome have been linked to many complex diseases. Similar to how genome-wide association studies (GWAS) have helped discover single-nucleotide polymorphisms linked to disease phenotypes, the extension of GWAS to CNVs has aided the discovery of structural variants associated with human traits and diseases. RESULTS: We present CoNVaQ, an easy-to-use web-based tool for CNV-based association studies. The web service allows users to upload two sets of CNV segments and search for genomic regions where the occurrence of CNVs is significantly associated with the phenotype. CoNVaQ provides two models: a simple statistical model using Fisher's exact test and a novel query-based model matching regions to user-defined queries. For each region, the method computes a global q-value statistic by repeated permutation of samples among the populations. We demonstrate our platform by using it to analyze a data set of HPV-positive and HPV-negative penile cancer patients. CONCLUSIONS: CoNVaQ provides a simple workflow for performing CNV-based association studies. It is made available as a web platform in order to provide a user-friendly workflow for biologists and clinicians to carry out CNV data analysis without installing any software. Through the web interface, users are also able to analyze their results to find overrepresented GO terms and pathways. In addition, our method is also available as a package for the R programming language. CoNVaQ is available at https://convaq.compbio.sdu.dk .