A systematic assessment of the impact of rare canonical splice site variants on splicing using functional and in silico methods.

Canonical splice site variants (CSSVs) are often presumed to cause loss-of-function (LoF) and are assigned very strong evidence of pathogenicity (according to American College of Medical Genetics/Association for Molecular Pathology criterion PVS1). The exact nature and predictability of splicing effects of unselected rare CSSVs in blood-expressed genes are poorly understood. We identified 168 rare CSSVs in blood-expressed genes in 112 individuals using genome sequencing, and studied their impact on splicing using RNA sequencing (RNA-seq). There was no evidence of a frameshift, nor of reduced expression consistent with nonsense-mediated decay, for 25.6% of CSSVs: 17.9% had wildtype splicing only and normal junction depths, 3.6% resulted in cryptic splice site usage and in-frame insertions or deletions, 3.6% resulted in full exon skipping (in frame), and 0.6% resulted in full intron inclusion (in frame). Blind to these RNA-seq data, we attempted to predict the precise impact of CSSVs by applying in silico tools and the ClinGen Sequence Variant Interpretation Working Group 2018 guidelines for applying PVS1 criterion. The predicted impact on splicing using (1) SpliceAI, (2) MaxEntScan, and (3) AutoPVS1, an automatic classification tool for PVS1 interpretation of null variants that utilizes Ensembl Variant Effect Predictor and MaxEntScan, was concordant with RNA-seq analyses for 65%, 63%, and 61% of CSSVs, respectively. In summary, approximately one in four rare CSSVs did not show evidence for LoF based on analysis of RNA-seq data. Predictions from in silico methods were often discordant with findings from RNA-seq. More caution may be warranted in applying PVS1-level evidence to CSSVs in the absence of functional data.

SEPT-GD: A decision tree to prioritise potential RNA splice variants in cardiomyopathy genes for functional splicing assays in diagnostics.

BACKGROUND: Splice prediction algorithms currently used in routine DNA diagnostics have limited sensitivity and specificity, therefore many potential splice variants are classified as variants of uncertain significance (VUSs). However, functional assessment of VUSs to test splicing is labour-intensive and time-consuming. We developed a decision tree to prioritise potential splice variants for functional studies and functionally verified the outcome of the decision tree. MATERIALS AND METHODS: We built the decision tree, SEPT-GD, by setting thresholds for the splice prediction programs implemented in Alamut. A set of 343 variants with known effects on splicing was used as control for sensitivity and specificity. We tested SEPT-GD using variants from a Dutch cardiomyopathy cohort of 2002 patients that were previously classified as VUS and predicted to have a splice effect according to diagnostic rules. We then selected 12 VUSs ranked by SEPT-GD to functionally verify the predicted effect on splicing using a minigene assay: 10 variants predicted to have a strong effect and 2 with a weak effect. RT-PCR was performed for nine variants. Variant classification was re-evaluated based on the functional test outcome. RESULTS: Compared to similar individually tested algorithms, SEPT-GD shows higher sensitivity (91 %) and comparable specificity (88 %) for both consensus (dinucleotides at the start and end of the intron, GT at the 5' end and AG at the 3' end) and non-consensus splice-site variants (excluding middle of exon variants). Using clinical diagnostic criteria, 1295 unique variants in our cardiomyopathy cohort had originally been classified as VUSs, with 57 predicted by Alamut to have an effect on splicing. Using SEPT-GD, we prioritised 31 variants in 40 patients. In the minigene assay, all 12 variants showed results concordant with SEPT-GD predictions. RT-PCR confirmed the minigene results for two variants, TMEM43 c.1000 + 5G > T and TTN c.25922-6 T > G. Based on all outcomes, the SGCD c.4-1G > A and CSRP3 c.282-5_285del variants were reclassified as likely pathogenic. CONCLUSION: SEPT-GD outperforms the tools commonly used for RNA splicing prediction and improves prioritisation of variants in cardiomyopathy genes for functional splicing analysis in a diagnostic setting.

Functional assessment of somatic STK11 variants identified in primary human non-small cell lung cancers.

Serine/Threonine Kinase 11 (STK11) encodes an important tumor suppressor that is frequently mutated in lung adenocarcinoma. Clinical studies have shown that mutations in STK11 resulting in loss of function correlate with resistance to anti-PD-1 monoclonal antibody therapy in KRAS-driven non-small cell lung cancer (NSCLC), but the molecular mechanisms responsible remain unclear. Despite this uncertainty, STK11 functional status is emerging as a reliable biomarker for predicting non-response to anti-PD-1 therapy in NSCLC patients. The clinical utility of this biomarker ultimately depends upon accurate classification of STK11 variants. For nonsense variants occurring early in the STK11 coding region, this assessment is straightforward. However, rigorously demonstrating the functional impact of missense variants remains an unmet challenge. Here we present data characterizing four STK11 splice-site variants by analyzing tumor mRNA, and 28 STK11 missense variants using an in vitro kinase assay combined with a cell-based p53-dependent luciferase reporter assay. The variants we report were identified in primary human NSCLC biopsies in collaboration with the University of Vermont Genomic Medicine group. Additionally, we compare our experimental results with data from 22 in silico predictive algorithms. Our work highlights the power, utility and necessity of functional variant assessment and will aid STK11 variant curation, provide a platform to assess novel STK11 variants and help guide anti-PD-1 therapy utilization in KRAS-driven NSCLCs.

Molecular Inversion Probe-Based Sequencing of USH2A Exons and Splice Sites as a Cost-Effective Screening Tool in USH2 and arRP Cases.

A substantial proportion of subjects with autosomal recessive retinitis pigmentosa (arRP) or Usher syndrome type II (USH2) lacks a genetic diagnosis due to incomplete USH2A screening in the early days of genetic testing. These cases lack eligibility for optimal genetic counseling and future therapy. USH2A defects are the most frequent cause of USH2 and are also causative in individuals with arRP. Therefore, USH2A is an important target for genetic screening. The aim of this study was to assess unscreened or incompletely screened and unexplained USH2 and arRP cases for (likely) pathogenic USH2A variants. Molecular inversion probe (MIP)-based sequencing was performed for the USH2A exons and their flanking regions, as well as published deep-intronic variants. This was done to identify single nucleotide variants (SNVs) and copy number variants (CNVs) in 29 unscreened or partially pre-screened USH2 and 11 partially pre-screened arRP subjects. In 29 out of these 40 cases, two (likely) pathogenic variants were successfully identified. Four of the identified SNVs and one CNV were novel. One previously identified synonymous variant was demonstrated to affect pre-mRNA splicing. In conclusion, genetic diagnoses were obtained for a majority of cases, which confirms that MIP-based sequencing is an effective screening tool for USH2A. Seven unexplained cases were selected for future analysis with whole genome sequencing.

Isoform-specific functions of an evolutionarily conserved 3 bp micro-exon alternatively spliced from another exon in Drosophila homothorax gene.

Micro-exons are exons of very small size (usually 3-30 nts). Some micro-exons are alternatively spliced. Their functions, regulation and evolution are largely unknown. Here, we present an example of an alternatively spliced 3 bp micro-exon (micro-Ex8) in the homothorax (hth) gene in Drosophila. Hth is involved in many developmental processes. It contains a MH domain and a TALE-class homeodomain (HD). It binds to another homeodomain Exd via its MH domain to promote the nuclear import of the Hth-Exd complex and serve as a cofactor for Hox proteins. The MH and HD domains in Hth as well as the HTh-Exd interaction are highly conserved in evolution. The alternatively spliced micro-exon lies between the exons encoding the MH and HD domains. We provide clear proof that the micro-Ex8 is produced by alternative splicing from a 48 bp full-length exon 8 (FL-Ex8) and the micro-Ex8 is the first three nt is FL-Ex8. We found that the micro-Ex8 is the ancient form and the 3 + 48 organization of alternatively spliced overlapping exons only emerged in the Schizophora group of Diptera and is absolutely conserved in this group. We then used several strategies to test the in vivo function of the two types of isoforms and found that the micro-Ex8 and FL-Ex8 isoforms have largely overlapping functions but also have non-redundant functions that are tissue-specific, which supports their strong evolutionary conservation. Since the different combinations of protein interaction of Hth with Exd and/or Hox can have different DNA target specificity, our finding of alternatively spliced isoforms adds to the spectrum of structural and functional diversity under developmental regulation.

Assessment of branch point prediction tools to predict physiological branch points and their alteration by variants.

BACKGROUND: Branch points (BPs) map within short motifs upstream of acceptor splice sites (3'ss) and are essential for splicing of pre-mature mRNA. Several BP-dedicated bioinformatics tools, including HSF, SVM-BPfinder, BPP, Branchpointer, LaBranchoR and RNABPS were developed during the last decade. Here, we evaluated their capability to detect the position of BPs, and also to predict the impact on splicing of variants occurring upstream of 3'ss. RESULTS: We used a large set of constitutive and alternative human 3'ss collected from Ensembl (n = 264,787 3'ss) and from in-house RNAseq experiments (n = 51,986 3'ss). We also gathered an unprecedented collection of functional splicing data for 120 variants (62 unpublished) occurring in BP areas of disease-causing genes. Branchpointer showed the best performance to detect the relevant BPs upstream of constitutive and alternative 3'ss (99.48 and 65.84% accuracies, respectively). For variants occurring in a BP area, BPP emerged as having the best performance to predict effects on mRNA splicing, with an accuracy of 89.17%. CONCLUSIONS: Our investigations revealed that Branchpointer was optimal to detect BPs upstream of 3'ss, and that BPP was most relevant to predict splicing alteration due to variants in the BP area.

The alternative splicing of SKU5-Similar3 in Arabidopsis.

Alternative splicing largely enhanced the diversity of transcriptome and proteome in eukaryas. Along with technological development, more and more genes are reported to be alternatively spliced during mRNA maturation. Here, I report the alternative splicing of SKU5-Similar 3 (SKS3) and its special splicing site in Arabidopsis. SKS3 was predicted to be alternatively transcribed into two variants, SKS3.1 and SKS3.2, which encoded a GPI-anchored protein and a soluble secretory protein, respectively. But, according to experimental data, instead of SKS3.2, a novel variant, SKS3.3, which encodes a protein with a transmembrane region at its C-terminus, was demonstrated. Interestingly, it exhibites a different organ-specific expression pattern with SKS3.1, and an unusual intron splicing site not following 'GT-AG' rule or any reported rule.

Mathematical modeling identifies potential gene structure determinants of co-transcriptional control of alternative pre-mRNA splicing.

The spliceosome catalyzes the removal of introns from pre-messenger RNA (mRNA) and subsequent pairing of exons with remarkable fidelity. Some exons are known to be skipped or included in the mature mRNA in a cell type- or context-dependent manner (cassette exons), thereby contributing to the diversification of the human proteome. Interestingly, splicing is initiated (and sometimes completed) co-transcriptionally. Here, we develop a kinetic mathematical modeling framework to investigate alternative co-transcriptional splicing (CTS) and, specifically, the control of cassette exons' inclusion. We show that when splicing is co-transcriptional, default splice patterns of exon inclusion are more likely than when splicing is post-transcriptional, and that certain exons are more likely to be regulatable (i.e. cassette exons) than others, based on the exon-intron structure context. For such regulatable exons, transcriptional elongation rates may affect splicing outcomes. Within the CTS paradigm, we examine previously described hypotheses of co-operativity between splice sites of short introns (i.e. 'intron definition') or across short exons (i.e. 'exon definition'), and find that models encoding these faithfully recapitulate observations in the fly and human genomes, respectively.

Markovian encoding models in human splice site recognition using SVM.

Splice site recognition is among the most significant and challenging tasks in bioinformatics due to its key role in gene annotation. Effective prediction of splice site requires nucleotide encoding methods that reveal the characteristics of DNA sequences to provide appropriate features to serve as input of machine learning classifiers. Markovian models are the most influential encoding methods that highly used for pattern recognition in biological data. However, a direct performance comparison of these methods in splice site domain has not been assessed yet. This study compares various Markovian encoding models for splice site prediction utilizing support vector machine, as the most outstanding learning method in the domain, and conducts a new precise evaluation of Markovian approaches that corrects this limitation. Moreover, a novel sequence encoding approach based on third order Markov model (MM3) is proposed. The experimental results show that the proposed method, namely MM3-SVM, performs significantly better than thirteen best known state-of-the-art algorithms, while tested on HS3D dataset considering several performance criteria. Further, it achieved higher prediction accuracy than several well-known tools like NNsplice, MEM, MM1, WMM, and GeneID, using an independent test set of 50 genes. We also developed MMSVM, a web tool to predict splice sites in any human sequence using the proposed approach. The MMSVM web server can be assessed at https://pashaei.shinyapps.io/mmsvm.

EYS mutation update: In silico assessment of 271 reported and 26 novel variants in patients with retinitis pigmentosa.

Mutations in Eyes shut homolog (EYS) are one of the most common causes of autosomal recessive (ar) retinitis pigmentosa (RP), a progressive blinding disorder. The exact function of the EYS protein and the pathogenic mechanisms underlying EYS-associated RP are still poorly understood, which hampers the interpretation of the causality of many EYS variants discovered to date. We collected all reported EYS variants present in 377 arRP index cases published before June 2017, and uploaded them in the Leiden Open Variation Database (www.LOVD.nl/EYS). We also describe 36 additional index cases, carrying 26 novel variants. Of the 297 unique EYS variants identified, almost half (n = 130) are predicted to result in premature truncation of the EYS protein. Classification of all variants using the American College of Medical Genetics and Genomics guidelines revealed that the predicted pathogenicity of these variants cover the complete spectrum ranging from likely benign to pathogenic, although especially missense variants largely fall in the category of uncertain significance. Besides the identification of likely benign alleles previously reported as being probably pathogenic, our comprehensive analysis underscores the need of functional assays to assess the causality of EYS variants, in order to improve molecular diagnostics and counseling of patients with EYS-associated RP.

A novel method for splice sites prediction using sequence component and hidden Markov model.

With increasing growth of DNA sequence data, it has become an urgent demand to develop new methods to accurately predict the genes. The performance of gene detection methods mainly depend on the efficiency of splice site prediction methods. In this paper, a novel method for detecting splice sites is proposed by using a new effective DNA encoding method and AdaBoost.M1 classifier. Our proposed DNA encoding method is based on multi-scale component (MSC) and first order Markov model (MM1). It has been applied to the HS3D dataset with repeated 10 fold cross validation. The experimental results indicate that the new method has increased the classification accuracy and outperformed some current methods such as MM1-SVM, Reduced MM1-SVM, SVM-B, LVMM, DM-SVM, DM2-AdaBoost and MS C+Pos(+APR)-SVM.

Single Molecule Cluster Analysis dissects splicing pathway conformational dynamics.

We report Single Molecule Cluster Analysis (SiMCAn), which utilizes hierarchical clustering of hidden Markov modeling-fitted single-molecule fluorescence resonance energy transfer (smFRET) trajectories to dissect the complex conformational dynamics of biomolecular machines. We used this method to study the conformational dynamics of a precursor mRNA during the splicing cycle as carried out by the spliceosome. By clustering common dynamic behaviors derived from selectively blocked splicing reactions, SiMCAn was able to identify the signature conformations and dynamic behaviors of multiple ATP-dependent intermediates. In addition, it identified an open conformation adopted late in splicing by a 3' splice-site mutant, invoking a mechanism for substrate proofreading. SiMCAn enables rapid interpretation of complex single-molecule behaviors and should prove useful for the comprehensive analysis of a plethora of dynamic cellular machines.

Molecular characterization of ten F8 splicing mutations in RNA isolated from patient's leucocytes: assessment of in silico prediction tools accuracy.

Although 8% of reported FVIII gene (F8) mutations responsible for haemophilia A (HA) affect mRNA processing, very few have been fully characterized at the mRNA level and/or systematically predicted their biological consequences by in silico analysis. This study is aimed to elucidate the effect of potential splice site mutations (PSSM) on the F8 mRNA processing, investigate its correlation with disease severity, and assess their concordance with in silico predictions. We studied the F8 mRNA from 10 HA patient's leucocytes with PSSM by RT-PCR and compared the experimental results with those predicted in silico. The mRNA analysis could explain all the phenotypes observed and demonstrated exon skipping in six cases (c.222G>A, c.601+1delG, c.602-11T>G, c.671-3C>G, c.6115+9C>G and c.6116-1G>A) and activation of cryptic splicing sites, both donor (c.1009+1G>A and c.1009+3A>C) and acceptor sites (c.266-3delC and c.5587-1G>A). In contrast, the in silico analysis was able to predict the score variation of most of the affected splice site, but the precise mechanism could only be correctly determined in two of the 10 mutations analysed. In addition, we have detected aberrant F8 transcripts, even in healthy controls, so this must be taken into account as they could mask the actual contribution of some PSSM. We conclude that F8 mRNA analysis using leucocytes still constitutes an excellent approach to investigate the transcriptional effects of the PSSM in HA, whereas prediction in silico is not always reliable for diagnostic decision-making.

Functional assessment of potential splice site variants in arrhythmogenic right ventricular dysplasia/cardiomyopathy.

BACKGROUND: Interpretation of genetic screening results in arrhythmogenic right ventricular dysplasia/cardiomyopathy (ARVD/C) often is difficult. Pathogenicity of variants with uncertain clinical significance may be predicted by software algorithms. However, functional assessment can unambiguously demonstrate the effect of such variants. OBJECTIVE: The purpose of this study was to perform functional analysis of potential splice site variants in ARVD/C patients. METHODS: Nine variants in desmosomal (PKP2, JUP, DSG2, DSC2) genes with potential RNA splicing effect were analyzed. The variants were found in patients who fulfilled 2010 ARVD/C Task Force Criteria (n = 7) or had suspected ARVD/C (n = 2). Total RNA was isolated from fresh blood samples and subjected to reverse transcriptase polymerase chain reaction. RESULTS: An effect on splicing was predicted by software algorithms for all variants. Of the 9 variants, 5 were intronic and 4 exonic. RNA analysis showed a functional effect on mRNA splicing by exon skipping, generation of new splice sites, or activation of cryptic sites in 6 variants. All 5 intronic variants tested severely impaired splicing. Only 1 of 4 exonic potential splice site variants was shown to have a deleterious effect on splicing. The remaining 3 exonic variants had no detectable effect on splicing, and heterozygous presence in mRNA confirmed biallelic expression. CONCLUSION: Six variants of uncertain clinical significance in the PKP2, JUP, and DSG2 genes showed a deleterious effect on mRNA splicing, indicating these are ARVD/C-related pathogenic splice site mutations. These results highlight the importance of functional assessment of potential splice site variants to improve patient care and facilitate cascade screening.

Experimental assessment of splicing variants using expression minigenes and comparison with in silico predictions.

Assessment of the functional consequences of variants near splice sites is a major challenge in the diagnostic laboratory. To address this issue, we created expression minigenes (EMGs) to determine the RNA and protein products generated by splice site variants (n = 10) implicated in cystic fibrosis (CF). Experimental results were compared with the splicing predictions of eight in silico tools. EMGs containing the full-length Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) coding sequence and flanking intron sequences generated wild-type transcript and fully processed protein in Human Embryonic Kidney (HEK293) and CF bronchial epithelial (CFBE41o-) cells. Quantification of variant induced aberrant mRNA isoforms was concordant using fragment analysis and pyrosequencing. The splicing patterns of c.1585-1G>A and c.2657+5G>A were comparable to those reported in primary cells from individuals bearing these variants. Bioinformatics predictions were consistent with experimental results for 9/10 variants (MES), 8/10 variants (NNSplice), and 7/10 variants (SSAT and Sroogle). Programs that estimate the consequences of mis-splicing predicted 11/16 (HSF and ASSEDA) and 10/16 (Fsplice and SplicePort) experimentally observed mRNA isoforms. EMGs provide a robust experimental approach for clinical interpretation of splice site variants and refinement of in silico tools.

Hereditary breast and ovarian cancer: assessment of point mutations and copy number variations in Brazilian patients.

BACKGROUND: Germ line mutations in BRCA1 and BRCA2 (BRCA1/2) and other susceptibility genes have been identified as genetic causes of hereditary breast and ovarian cancer (HBOC). To identify the disease-causing mutations in a cohort of 120 Brazilian women fulfilling criteria for HBOC, we carried out a comprehensive screening of BRCA1/2, TP53 R337H, CHEK2 1100delC, followed by an analysis of copy number variations in 14 additional breast cancer susceptibility genes (PTEN, ATM, NBN, RAD50, RAD51, BRIP1, PALB2, MLH1, MSH2, MSH6, TP53, CDKN2A, CDH1 and CTNNB1). METHODS: Capillary sequencing and multiplex ligation-dependent probe amplification (MLPA) were used for detecting point mutations and copy number variations (CNVs), respectively, for the BRCA1 and BRCA2 genes; capillary sequencing was used for point mutation for both variants TP53 R337H and CHEK2 1100delC, and finally array comparative genomic hybridization (array-CGH) was used for identifying CNVs in the 14 additional genes. RESULTS: The positive detection rate in our series was 26%. BRCA1 pathogenic mutations were found in 20 cases, including two cases with CNVs, whereas BRCA2 mutations were found in 7 cases. We also found three patients with the TP53 R337H mutation and one patient with the CHEK2 1100delC mutation. Seven (25%) pathogenic mutations in BRCA1/2 were firstly described, including a splice-site BRCA1 mutation for which pathogenicity was confirmed by the presence of an aberrant transcript showing the loss of the last 62 bp of exon 7. Microdeletions of exon 4 in ATM and exon 2 in PTEN were identified in BRCA2-mutated and BRCA1/2-negative patients, respectively. CONCLUSIONS: In summary, our results showed a high frequency of BRCA1/2 mutations and a higher prevalence of BRCA1 (64.5%) gene. Moreover, the detection of the TP53 R337H variant in our series and the fact that this variant has a founder effect in our population prompted us to suggest that all female breast cancer patients with clinical criteria for HBOC and negative for BRCA1/2 genes should be tested for the TP53 R337H variant. Furthermore, the presence of genomic structural rearrangement resulting in CNVs in other genes that predispose breast cancer in conjunction with BRCA2 point mutations demonstrated a highly complex genetic etiology in Brazilian breast cancer families.

FineSplice, enhanced splice junction detection and quantification: a novel pipeline based on the assessment of diverse RNA-Seq alignment solutions.

Alternative splicing is the main mechanism governing protein diversity. The recent developments in RNA-Seq technology have enabled the study of the global impact and regulation of this biological process. However, the lack of standardized protocols constitutes a major bottleneck in the analysis of alternative splicing. This is particularly important for the identification of exon-exon junctions, which is a critical step in any analysis workflow. Here we performed a systematic benchmarking of alignment tools to dissect the impact of design and method on the mapping, detection and quantification of splice junctions from multi-exon reads. Accordingly, we devised a novel pipeline based on TopHat2 combined with a splice junction detection algorithm, which we have named FineSplice. FineSplice allows effective elimination of spurious junction hits arising from artefactual alignments, achieving up to 99% precision in both real and simulated data sets and yielding superior F1 scores under most tested conditions. The proposed strategy conjugates an efficient mapping solution with a semi-supervised anomaly detection scheme to filter out false positives and allows reliable estimation of expressed junctions from the alignment output. Ultimately this provides more accurate information to identify meaningful splicing patterns. FineSplice is freely available at https://sourceforge.net/p/finesplice/.

Assessment of transcript reconstruction methods for RNA-seq.

We evaluated 25 protocol variants of 14 independent computational methods for exon identification, transcript reconstruction and expression-level quantification from RNA-seq data. Our results show that most algorithms are able to identify discrete transcript components with high success rates but that assembly of complete isoform structures poses a major challenge even when all constituent elements are identified. Expression-level estimates also varied widely across methods, even when based on similar transcript models. Consequently, the complexity of higher eukaryotic genomes imposes severe limitations on transcript recall and splice product discrimination that are likely to remain limiting factors for the analysis of current-generation RNA-seq data.

A comprehensive, sensitive and economical approach for the detection of mutations in the RB1 gene in retinoblastoma.

Retinoblastoma (Rb) is the most common primary intraocular malignancy in children. It is brought about by the mutational inactivation of both alleles of RB1 gene in the developing retina. To identify the RB1 mutations, we analysed 74 retinoblastoma patients by screening the exons and the promoter region of RB1. The strategy used was to detect large deletions/duplications by fluorescent quantitative multiplex PCR; small deletions/insertions by fluorescent genotyping of RB1 alleles, and point mutations by PCR-RFLP and sequencing. Genomic DNA from the peripheral blood leucocytes of 74 Rb patients (53 with bilateral Rb, 21 with unilateral Rb; 4 familial cases) was screened for mutations. Recurrent mutations were identified in five patients with bilateral Rb, large deletions in 11 patients (nine with bilateral Rb and two with unilateral Rb), small deletions/insertions were found in 12 patients all with bilateral Rb, and point mutations in 26 patients (14 nonsense, six splice site, five substitution and one silent change). Three mutations were associated with variable expressivity of the disease in different family members. Using this method, the detection rates achieved in patients with bilateral Rb were 44/53 (83%) and with unilateral Rb, 5/21 (23.8%). This approach may be feasible for clinical genetic testing and counselling of patients.

Accurate splice site prediction using support vector machines.

BACKGROUND: For splice site recognition, one has to solve two classification problems: discriminating true from decoy splice sites for both acceptor and donor sites. Gene finding systems typically rely on Markov Chains to solve these tasks. RESULTS: In this work we consider Support Vector Machines for splice site recognition. We employ the so-called weighted degree kernel which turns out well suited for this task, as we will illustrate in several experiments where we compare its prediction accuracy with that of recently proposed systems. We apply our method to the genome-wide recognition of splice sites in Caenorhabditis elegans, Drosophila melanogaster, Arabidopsis thaliana, Danio rerio, and Homo sapiens. Our performance estimates indicate that splice sites can be recognized very accurately in these genomes and that our method outperforms many other methods including Markov Chains, GeneSplicer and SpliceMachine. We provide genome-wide predictions of splice sites and a stand-alone prediction tool ready to be used for incorporation in a gene finder. AVAILABILITY: Data, splits, additional information on the model selection, the whole genome predictions, as well as the stand-alone prediction tool are available for download at http://www.fml.mpg.de/raetsch/projects/splice.