Functional Copy-Number Alterations as Diagnostic and Prognostic Biomarkers in Neuroendocrine Tumors.

Functional copy-number alterations (fCNAs) are DNA copy-number changes with concordant differential gene expression. These are less likely to be bystander genetic lesions and could serve as robust and reproducible tumor biomarkers. To identify candidate fCNAs in neuroendocrine tumors (NETs), we integrated chromosomal microarray (CMA) and RNA-seq differential gene-expression data from 31 pancreatic (pNETs) and 33 small-bowel neuroendocrine tumors (sbNETs). Tumors were resected from 47 early-disease-progression (<24 months) and 17 late-disease-progression (>24 months) patients. Candidate fCNAs that accurately differentiated these groups in this discovery cohort were then replicated using fluorescence in situ hybridization (FISH) on formalin-fixed, paraffin-embedded (FFPE) tissues in a larger validation cohort of 60 pNETs and 82 sbNETs (52 early- and 65 late-disease-progression samples). Logistic regression analysis revealed the predictive ability of these biomarkers, as well as the assay-performance metrics of sensitivity, specificity, and area under the curve. Our results indicate that copy-number changes at chromosomal loci 4p16.3, 7q31.2, 9p21.3, 17q12, 18q21.2, and 19q12 may be used as diagnostic and prognostic NET biomarkers. This involves a rapid, cost-effective approach to determine the primary tumor site for patients with metastatic liver NETs and to guide risk-stratified therapeutic decisions.

Assessing variants of uncertain significance implicated in hearing loss using a comprehensive deafness proteome.

Hearing loss is the leading sensory deficit, affecting ~ 5% of the population. It exhibits remarkable heterogeneity across 223 genes with 6328 pathogenic missense variants, making deafness-specific expertise a prerequisite for ascribing phenotypic consequences to genetic variants. Deafness-implicated variants are curated in the Deafness Variation Database (DVD) after classification by a genetic hearing loss expert panel and thorough informatics pipeline. However, seventy percent of the 128,167 missense variants in the DVD are "variants of uncertain significance" (VUS) due to insufficient evidence for classification. Here, we use the deep learning protein prediction algorithm, AlphaFold2, to curate structures for all DVD genes. We refine these structures with global optimization and the AMOEBA force field and use DDGun3D to predict folding free energy differences (∆∆GFold) for all DVD missense variants. We find that 5772 VUSs have a large, destabilizing ∆∆GFold that is consistent with pathogenic variants. When also filtered for CADD scores (> 25.7), we determine 3456 VUSs are likely pathogenic at a probability of 99.0%. Of the 224 genes in the DVD, 166 genes (74%) exhibit one or more missense variants predicted to cause a pathogenic change in protein folding stability. The VUSs prioritized here affect 119 patients (~ 3% of cases) sequenced by the OtoSCOPE targeted panel. Approximately half of these patients previously received an inconclusive report, and reclassification of these VUSs as pathogenic provides a new genetic diagnosis for six patients.

Prioritization of Fluorescence In Situ Hybridization (FISH) Probes for Differentiating Primary Sites of Neuroendocrine Tumors with Machine Learning.

Determining neuroendocrine tumor (NET) primary sites is pivotal for patient care as pancreatic NETs (pNETs) and small bowel NETs (sbNETs) have distinct treatment approaches. The diagnostic power and prioritization of fluorescence in situ hybridization (FISH) assay biomarkers for establishing primary sites has not been thoroughly investigated using machine learning (ML) techniques. We trained ML models on FISH assay metrics from 85 sbNET and 59 pNET samples for primary site prediction. Exploring multiple methods for imputing missing data, the impute-by-median dataset coupled with a support vector machine model achieved the highest classification accuracy of 93.1% on a held-out test set, with the top importance variables originating from the ERBB2 FISH probe. Due to the greater interpretability of decision tree (DT) models, we fit DT models to ten dataset splits, achieving optimal performance with k-nearest neighbor (KNN) imputed data and a transformation to single categorical biomarker probe variables, with a mean accuracy of 81.4%, on held-out test sets. ERBB2 and MET variables ranked as top-performing features in 9 of 10 DT models and the full dataset model. These findings offer probabilistic guidance for FISH testing, emphasizing the prioritization of the ERBB2, SMAD4, and CDKN2A FISH probes in diagnosing NET primary sites.

A splice site mutation in the TSEN2 causes a new syndrome with craniofacial and central nervous system malformations, and atypical hemolytic uremic syndrome.

Recessive mutations in the genes encoding the four subunits of the tRNA splicing endonuclease complex (TSEN54, TSEN34, TSEN15, and TSEN2) cause various forms of pontocerebellar hypoplasia, a disorder characterized by hypoplasia of the cerebellum and the pons, microcephaly, dysmorphisms, and other variable clinical features. Here, we report an intronic recessive founder variant in the gene TSEN2 that results in abnormal splicing of the mRNA of this gene, in six individuals from four consanguineous families affected with microcephaly, multiple craniofacial malformations, radiological abnormalities of the central nervous system, and cognitive retardation of variable severity. Remarkably, unlike patients with previously described mutations in the components of the TSEN complex, all the individuals that we report developed atypical hemolytic uremic syndrome (aHUS) with thrombotic microangiopathy, microangiopathic hemolytic anemia, thrombocytopenia, proteinuria, severe hypertension, and end-stage kidney disease (ESKD) early in life. Bulk RNA sequencing of peripheral blood cells of four affected individuals revealed abnormal tRNA transcripts, indicating an alteration of the tRNA biogenesis. Morpholino-mediated skipping of exon 10 of tsen2 in zebrafish produced phenotypes similar to human patients. Thus, we have identified a novel syndrome accompanied by aHUS suggesting the existence of a link between tRNA biology and vascular endothelium homeostasis, which we propose to name with the acronym TRACK syndrome (TSEN2 Related Atypical hemolytic uremic syndrome, Craniofacial malformations, Kidney failure).

Genomic Landscape and Mutational Signatures of Deafness-Associated Genes.

The classification of genetic variants represents a major challenge in the post-genome era by virtue of their extraordinary number and the complexities associated with ascribing a clinical impact, especially for disorders exhibiting exceptional phenotypic, genetic, and allelic heterogeneity. To address this challenge for hearing loss, we have developed the Deafness Variation Database (DVD), a comprehensive, open-access resource that integrates all available genetic, genomic, and clinical data together with expert curation to generate a single classification for each variant in 152 genes implicated in syndromic and non-syndromic deafness. We evaluate 876,139 variants and classify them as pathogenic or likely pathogenic (more than 8,100 variants), benign or likely benign (more than 172,000 variants), or of uncertain significance (more than 695,000 variants); 1,270 variants are re-categorized based on expert curation and in 300 instances, the change is of medical significance and impacts clinical care. We show that more than 96% of coding variants are rare and novel and that pathogenicity is driven by minor allele frequency thresholds, variant effect, and protein domain. The mutational landscape we define shows complex gene-specific variability, making an understanding of these nuances foundational for improved accuracy in variant interpretation in order to enhance clinical decision making and improve our understanding of deafness biology.

Presacral neuroendocrine tumors associated with the Currarino syndrome.

Currarino syndrome (CS) is an autosomal dominant syndrome caused by mutations in MNX1 and characterized by anorectal abnormalities, partial sacral agenesis, and presacral masses. The presacral masses are typically benign; however, malignant degeneration can occur, and presacral neuroendocrine tumors (NETs) have been reported in six cases. We report three individuals from two families affected by CS in which multiple individuals developed presacral NETs. The first family, 491, had six members with features of CS, including two siblings who presented with presacral, Grade 2 NETs, one of which had metastasized to bone and lymph nodes. A germline c.874C>T (p.Arg292Trp) mutation was found in a highly conserved region of MNX1 in three affected members who underwent sequencing. A second somatic variant/deletion in MNX1 was not detected in either patient's tumor. In the second family, 342, the proband presented with an incidentally discovered presacral NET. The proband's father had previously undergone resection of a presacral NET, and so genetic testing was performed, which did not reveal an MNX1 mutation or copy number variants. The lack of a second, somatic mutation in the tumors from family 491 argues against MNX1 acting as a tumor suppressor, and the absence of a germline MNX1 mutation in family 342 suggests that other genetic and anatomic factors contribute to the development of presacral NETs. These cases highlight the variable presentation of CS, and the potential for malignancy in these patients.

A comparative analysis of genetic hearing loss phenotypes in European/American and Japanese populations.

We present detailed comparative analyses to assess population-level differences in patterns of genetic deafness between European/American and Japanese cohorts with non-syndromic hearing loss. One thousand eighty-three audiometric test results (921 European/American and 162 Japanese) from members of 168 families (48 European/American and 120 Japanese) with non-syndromic hearing loss secondary to pathogenic variants in one of three genes (KCNQ4, TECTA, WFS1) were studied. Audioprofile characteristics, specific mutation types, and protein domains were considered in the comparative analyses. Our findings support differences in audioprofiles driven by both mutation type (non-truncating vs. truncating) and ethnic background. The former finding confirms data that ascribe a phenotypic consequence to different mutation types in KCNQ4; the latter finding suggests that there are ethnic-specific effects (genetic and/or environmental) that impact gene-specific audioprofiles for TECTA and WFS1. Identifying the drivers of ethnic differences will refine our understanding of phenotype-genotype relationships and the biology of hearing and deafness.

Structural Insights into Hearing Loss Genetics from Polarizable Protein Repacking.

Hearing loss is associated with ∼8100 mutations in 152 genes, and within the coding regions of these genes are over 60,000 missense variants. The majority of these variants are classified as "variants of uncertain significance" to reflect our inability to ascribe a phenotypic effect to the observed amino acid change. A promising source of pathogenicity information is biophysical simulation, although input protein structures often contain defects because of limitations in experimental data and/or only distant homology to a template. Here, we combine the polarizable atomic multipole optimized energetics for biomolecular applications force field, many-body optimization theory, and graphical processing unit acceleration to repack all deafness-associated proteins and thereby improve average structure MolProbity score from 2.2 to 1.0. We then used these optimized wild-type models to create over 60,000 structures for missense variants in the Deafness Variation Database, which are being incorporated into the Deafness Variation Database to inform deafness pathogenicity prediction. Finally, this work demonstrates that advanced polarizable atomic multipole force fields are efficient enough to repack the entire human proteome.

Machine learning with the TCGA-HNSC dataset: improving usability by addressing inconsistency, sparsity, and high-dimensionality.

BACKGROUND: In the era of precision oncology and publicly available datasets, the amount of information available for each patient case has dramatically increased. From clinical variables and PET-CT radiomics measures to DNA-variant and RNA expression profiles, such a wide variety of data presents a multitude of challenges. Large clinical datasets are subject to sparsely and/or inconsistently populated fields. Corresponding sequencing profiles can suffer from the problem of high-dimensionality, where making useful inferences can be difficult without correspondingly large numbers of instances. In this paper we report a novel deployment of machine learning techniques to handle data sparsity and high dimensionality, while evaluating potential biomarkers in the form of unsupervised transformations of RNA data. We apply preprocessing, MICE imputation, and sparse principal component analysis (SPCA) to improve the usability of more than 500 patient cases from the TCGA-HNSC dataset for enhancing future oncological decision support for Head and Neck Squamous Cell Carcinoma (HNSCC). RESULTS: Imputation was shown to improve prognostic ability of sparse clinical treatment variables. SPCA transformation of RNA expression variables reduced runtime for RNA-based models, though changes to classifier performance were not significant. Gene ontology enrichment analysis of gene sets associated with individual sparse principal components (SPCs) are also reported, showing that both high- and low-importance SPCs were associated with cell death pathways, though the high-importance gene sets were found to be associated with a wider variety of cancer-related biological processes. CONCLUSIONS: MICE imputation allowed us to impute missing values for clinically informative features, improving their overall importance for predicting two-year recurrence-free survival by incorporating variance from other clinical variables. Dimensionality reduction of RNA expression profiles via SPCA reduced both computation cost and model training/evaluation time without affecting classifier performance, allowing researchers to obtain experimental results much more quickly. SPCA simultaneously provided a convenient avenue for consideration of biological context via gene ontology enrichment analysis.

Correction to: Genomics of NSCLC patients both affirm PD-L1 expression and predict their clinical responses to anti-PD-1 immunotherapy.

It has been highlighted that in the original manuscript [1] Table S3 'An example of the predictive computational modeling process. Specific details on an annexure section of the PD-L1 pathway show the step-by-step reactions, mechanisms, and reaction equations that occur. Such reactions also occurred in all of the other pathways' was omitted and did not appear in the Additional files and that the Additional files were miss-numbered thereafter. This Correction shows the correct and incorrect Additional files. The original article has been updated.

Genomics of NSCLC patients both affirm PD-L1 expression and predict their clinical responses to anti-PD-1 immunotherapy.

BACKGROUND: Programmed Death Ligand 1 (PD-L1) is a co-stimulatory and immune checkpoint protein. PD-L1 expression in non-small cell lung cancers (NSCLC) is a hallmark of adaptive resistance and its expression is often used to predict the outcome of Programmed Death 1 (PD-1) and PD-L1 immunotherapy treatments. However, clinical benefits do not occur in all patients and new approaches are needed to assist in selecting patients for PD-1 or PD-L1 immunotherapies. Here, we hypothesized that patient tumor cell genomics influenced cell signaling and expression of PD-L1, chemokines, and immunosuppressive molecules and these profiles could be used to predict patient clinical responses. METHODS: We used a recent dataset from NSCLC patients treated with pembrolizumab. Deleterious gene mutational profiles in patient exomes were identified and annotated into a cancer network to create NSCLC patient-specific predictive computational simulation models. Validation checks were performed on the cancer network, simulation model predictions, and PD-1 match rates between patient-specific predicted and clinical responses. RESULTS: Expression profiles of these 24 chemokines and immunosuppressive molecules were used to identify patients who would or would not respond to PD-1 immunotherapy. PD-L1 expression alone was not sufficient to predict which patients would or would not respond to PD-1 immunotherapy. Adding chemokine and immunosuppressive molecule expression profiles allowed patient models to achieve a greater than 85.0% predictive correlation among predicted and reported patient clinical responses. CONCLUSIONS: Our results suggested that chemokine and immunosuppressive molecule expression profiles can be used to accurately predict clinical responses thus differentiating among patients who would and would not benefit from PD-1 or PD-L1 immunotherapies.

Utilizing ethnic-specific differences in minor allele frequency to recategorize reported pathogenic deafness variants.

Ethnic-specific differences in minor allele frequency impact variant categorization for genetic screening of nonsyndromic hearing loss (NSHL) and other genetic disorders. We sought to evaluate all previously reported pathogenic NSHL variants in the context of a large number of controls from ethnically distinct populations sequenced with orthogonal massively parallel sequencing methods. We used HGMD, ClinVar, and dbSNP to generate a comprehensive list of reported pathogenic NSHL variants and re-evaluated these variants in the context of 8,595 individuals from 12 populations and 6 ethnically distinct major human evolutionary phylogenetic groups from three sources (Exome Variant Server, 1000 Genomes project, and a control set of individuals created for this study, the OtoDB). Of the 2,197 reported pathogenic deafness variants, 325 (14.8%) were present in at least one of the 8,595 controls, indicating a minor allele frequency (MAF) > 0.00006. MAFs ranged as high as 0.72, a level incompatible with pathogenicity for a fully penetrant disease like NSHL. Based on these data, we established MAF thresholds of 0.005 for autosomal-recessive variants (excluding specific variants in GJB2) and 0.0005 for autosomal-dominant variants. Using these thresholds, we recategorized 93 (4.2%) of reported pathogenic variants as benign. Our data show that evaluation of reported pathogenic deafness variants using variant MAFs from multiple distinct ethnicities and sequenced by orthogonal methods provides a powerful filter for determining pathogenicity. The proposed MAF thresholds will facilitate clinical interpretation of variants identified in genetic testing for NSHL. All data are publicly available to facilitate interpretation of genetic variants causing deafness.

Clinically Focused Molecular Investigation of 1000 Consecutive Families with Inherited Retinal Disease.

PURPOSE: To devise a comprehensive multiplatform genetic testing strategy for inherited retinal disease and to describe its performance in 1000 consecutive families seen by a single clinician. DESIGN: Retrospective series. PARTICIPANTS: One thousand consecutive families seen by a single clinician. METHODS: The clinical records of all patients seen by a single retina specialist between January 2010 and June 2016 were reviewed, and all patients who met the clinical criteria for a diagnosis of inherited retinal disease were included in the study. Each patient was assigned to 1 of 62 diagnostic categories, and this clinical diagnosis was used to define the scope and order of the molecular investigations that were performed. The number of nucleotides evaluated in a given subject ranged from 2 to nearly 900 000. MAIN OUTCOME MEASURES: Sensitivity and false genotype rate. RESULTS: Disease-causing genotypes were identified in 760 families (76%). These genotypes were distributed across 104 different genes. More than 75% of these 104 genes have coding sequences small enough to be packaged efficiently into an adeno-associated virus. Mutations in ABCA4 were the most common cause of disease in this cohort (173 families), whereas mutations in 80 genes caused disease in 5 or fewer families (i.e., 0.5% or less). Disease-causing genotypes were identified in 576 of the families without next-generation sequencing (NGS). This included 23 families with mutations in the repetitive region of RPGR exon 15 that would have been missed by NGS. Whole-exome sequencing of the remaining 424 families revealed mutations in an additional 182 families, and whole-genome sequencing of 4 of the remaining 242 families revealed 2 additional genotypes that were invisible by the other methods. Performing the testing in a clinically focused tiered fashion would be 6.1% more sensitive and 17.7% less expensive and would have a significantly lower average false genotype rate than using whole-exome sequencing to assess more than 300 genes in all patients (7.1% vs. 128%; P < 0.001). CONCLUSIONS: Genetic testing for inherited retinal disease is now more than 75% sensitive. A clinically directed tiered testing strategy can increase sensitivity and improve statistical significance without increasing cost.

Heterozygous deep-intronic variants and deletions in ABCA4 in persons with retinal dystrophies and one exonic ABCA4 variant.

Variants in ABCA4 are responsible for autosomal-recessive Stargardt disease and cone-rod dystrophy. Sequence analysis of ABCA4 exons previously revealed one causative variant in each of 45 probands. To identify the "missing" variants in these cases, we performed multiplex ligation-dependent probe amplification-based deletion scanning of ABCA4. In addition, we sequenced the promoter region, fragments containing five deep-intronic splice variants, and 15 deep-intronic regions containing weak splice sites. Heterozygous deletions spanning ABCA4 exon 5 or exons 20-22 were found in two probands, heterozygous deep-intronic variants were identified in six probands, and a deep-intronic variant was found together with an exon 20-22 deletion in one proband. Based on ophthalmologic findings and characteristics of the identified exonic variants present in trans, the deep-intronic variants V1 and V4 were predicted to be relatively mild and severe, respectively. These findings are important for proper genetic counseling and for the development of variant-specific therapies.

Non-exomic and synonymous variants in ABCA4 are an important cause of Stargardt disease.

Mutations in ABCA4 cause Stargardt disease and other blinding autosomal recessive retinal disorders. However, sequencing of the complete coding sequence in patients with clinical features of Stargardt disease sometimes fails to detect one or both mutations. For example, among 208 individuals with clear clinical evidence of ABCA4 disease ascertained at a single institution, 28 had only one disease-causing allele identified in the exons and splice junctions of the primary retinal transcript of the gene. Haplotype analysis of these 28 probands revealed 3 haplotypes shared among ten families, suggesting that 18 of the 28 missing alleles were rare enough to be present only once in the cohort. We hypothesized that mutations near rare alternate splice junctions in ABCA4 might cause disease by increasing the probability of mis-splicing at these sites. Next-generation sequencing of RNA extracted from human donor eyes revealed more than a dozen alternate exons that are occasionally incorporated into the ABCA4 transcript in normal human retina. We sequenced the genomic DNA containing 15 of these minor exons in the 28 one-allele subjects and observed five instances of two different variations in the splice signals of exon 36.1 that were not present in normal individuals (P < 10(-6)). Analysis of RNA obtained from the keratinocytes of patients with these mutations revealed the predicted alternate transcript. This study illustrates the utility of RNA sequence analysis of human donor tissue and patient-derived cell lines to identify mutations that would be undetectable by exome sequencing.

Cordova: web-based management of genetic variation data.

UNLABELLED: Cordova is an out-of-the-box solution for building and maintaining an online database of genetic variations integrated with pathogenicity prediction results from popular algorithms. Our primary motivation for developing this system is to aid researchers and clinician-scientists in determining the clinical significance of genetic variations. To achieve this goal, Cordova provides an interface to review and manually or computationally curate genetic variation data as well as share it for clinical diagnostics and the advancement of research. AVAILABILITY AND IMPLEMENTATION: Cordova is open source under the MIT license and is freely available for download at https://github.com/clcg/cordova.

Calpain-5 mutations cause autoimmune uveitis, retinal neovascularization, and photoreceptor degeneration.

Autosomal dominant neovascular inflammatory vitreoretinopathy (ADNIV) is an autoimmune condition of the eye that sequentially mimics uveitis, retinitis pigmentosa, and proliferative diabetic retinopathy as it progresses to complete blindness. We identified two different missense mutations in the CAPN5 gene in three ADNIV kindreds. CAPN5 encodes calpain-5, a calcium-activated cysteine protease that is expressed in retinal photoreceptor cells. Both mutations cause mislocalization from the cell membrane to the cytosol, and structural modeling reveals that both mutations lie within a calcium-sensitive domain near the active site. CAPN5 is only the second member of the large calpain gene family to cause a human Mendelian disorder, and this is the first report of a specific molecular cause for autoimmune eye disease. Further investigation of these mutations is likely to provide insight into the pathophysiologic mechanisms of common diseases ranging from autoimmune disorders to diabetic retinopathy.

Transcriptomic analysis across nasal, temporal, and macular regions of human neural retina and RPE/choroid by RNA-Seq.

Proper spatial differentiation of retinal cell types is necessary for normal human vision. Many retinal diseases, such as Best disease and male germ cell associated kinase (MAK)-associated retinitis pigmentosa, preferentially affect distinct topographic regions of the retina. While much is known about the distribution of cell types in the retina, the distribution of molecular components across the posterior pole of the eye has not been well-studied. To investigate regional difference in molecular composition of ocular tissues, we assessed differential gene expression across the temporal, macular, and nasal retina and retinal pigment epithelium (RPE)/choroid of human eyes using RNA-Seq. RNA from temporal, macular, and nasal retina and RPE/choroid from four human donor eyes was extracted, poly-A selected, fragmented, and sequenced as 100 bp read pairs. Digital read files were mapped to the human genome and analyzed for differential expression using the Tuxedo software suite. Retina and RPE/choroid samples were clearly distinguishable at the transcriptome level. Numerous transcription factors were differentially expressed between regions of the retina and RPE/choroid. Photoreceptor-specific genes were enriched in the peripheral samples, while ganglion cell and amacrine cell genes were enriched in the macula. Within the RPE/choroid, RPE-specific genes were upregulated at the periphery while endothelium associated genes were upregulated in the macula. Consistent with previous studies, BEST1 expression was lower in macular than extramacular regions. The MAK gene was expressed at lower levels in macula than in extramacular regions, but did not exhibit a significant difference between nasal and temporal retina. The regional molecular distinction is greatest between macula and periphery and decreases between different peripheral regions within a tissue. Datasets such as these can be used to prioritize candidate genes for possible involvement in retinal diseases with regional phenotypes.

Advancing genetic testing for deafness with genomic technology.

BACKGROUND: Non-syndromic hearing loss (NSHL) is the most common sensory impairment in humans. Until recently its extreme genetic heterogeneity precluded comprehensive genetic testing. Using a platform that couples targeted genomic enrichment (TGE) and massively parallel sequencing (MPS) to sequence all exons of all genes implicated in NSHL, we tested 100 persons with presumed genetic NSHL and in so doing established sequencing requirements for maximum sensitivity and defined MPS quality score metrics that obviate Sanger validation of variants. METHODS: We examined DNA from 100 sequentially collected probands with presumed genetic NSHL without exclusions due to inheritance, previous genetic testing, or type of hearing loss. We performed TGE using post-capture multiplexing in variable pool sizes followed by Illumina sequencing. We developed a local Galaxy installation on a high performance computing cluster for bioinformatics analysis. RESULTS: To obtain maximum variant sensitivity with this platform 3.2-6.3 million total mapped sequencing reads per sample were required. Quality score analysis showed that Sanger validation was not required for 95% of variants. Our overall diagnostic rate was 42%, but this varied by clinical features from 0% for persons with asymmetric hearing loss to 56% for persons with bilateral autosomal recessive NSHL. CONCLUSIONS: These findings will direct the use of TGE and MPS strategies for genetic diagnosis for NSHL. Our diagnostic rate highlights the need for further research on genetic deafness focused on novel gene identification and an improved understanding of the role of non-exonic mutations. The unsolved families we have identified provide a valuable resource to address these areas.

Prioritization of retinal disease genes: an integrative approach.

The discovery of novel disease-associated variations in genes is often a daunting task in highly heterogeneous disease classes. We seek a generalizable algorithm that integrates multiple publicly available genomic data sources in a machine-learning model for the prioritization of candidates identified in patients with retinal disease. To approach this problem, we generate a set of feature vectors from publicly available microarray, RNA-seq, and ChIP-seq datasets of biological relevance to retinal disease, to observe patterns in gene expression specificity among tissues of the body and the eye, in addition to photoreceptor-specific signals by the CRX transcription factor. Using these features, we describe a novel algorithm, positive and unlabeled learning for prioritization (PULP). This article compares several popular supervised learning techniques as the regression function for PULP. The results demonstrate a highly significant enrichment for previously characterized disease genes using a logistic regression method. Finally, a comparison of PULP with the popular gene prioritization tool ENDEAVOUR shows superior prioritization of retinal disease genes from previous studies. The java source code, compiled binary, assembled feature vectors, and instructions are available online at https://github.com/ahwagner/PULP.