Distinguishing cancer-associated missense mutations from common polymorphisms

Incorporating molecular and functional context into the analysis and prioritization of human variants associated with cancer

BACKGROUND AND OBJECTIVE

With recent breakthroughs in high-throughput sequencing, identifying deleterious mutations is one of the key challenges for personalized medicine. At the gene and protein level, it has proven difficult to determine the impact of previously unknown variants. A statistical method has been developed to assess the significance of disease mutation clusters on protein domains by incorporating domain functional annotations to assist in the functional characterization of novel variants.

METHODS

Disease mutations aggregated from multiple databases were mapped to domains, and were classified as either cancer- or non-cancer-related. The statistical method for identifying significantly disease-associated domain positions was applied to both sets of mutations and to randomly generated mutation sets for comparison. To leverage the known function of protein domain regions, the method optionally distributes significant scores to associated functional feature positions.

RESULTS

Most disease mutations are localized within protein domains and display a tendency to cluster at individual domain positions. The method identified significant disease mutation hotspots in both the cancer and non-cancer datasets. The domain significance scores (DS-scores) for cancer form a bimodal distribution with hotspots in oncogenes forming a second peak at higher DS-scores than non-cancer, and hotspots in tumor suppressors have scores more similar to non-cancers. In addition, on an independent mutation benchmarking set, the DS-score method identified mutations known to alter protein function with very high precision.

CONCLUSION

By aggregating mutations with known disease association at the domain level, the method was able to discover domain positions enriched with multiple occurrences of deleterious mutations while incorporating relevant functional annotations. The method can be incorporated into translational bioinformatics tools to characterize rare and novel variants within large-scale sequencing studies.

Oncogenomics methods and resources

Today, cancer is viewed as a genetic disease and many genetic mechanisms of oncogenesis are known. The progression from normal tissue to invasive cancer is thought to occur over a timescale of 5-20 years. This transformation is driven by both inherited genetic factors and somatic genetic alterations and mutations, and it results in uncontrolled cell growth and, in many cases, death. In this article, we review the main types of genomic and genetic alterations involved in cancer, namely copy-number changes, genomic rearrangements, somatic mutations, polymorphisms, and epigenomic alterations in cancer. We then discuss the transcriptomic consequences of these alterations in tumor cells. The use of "next-generation" sequencing methods in cancer research is described in the relevant sections. Finally, we discuss different approaches for candidate prioritization and integration and analysis of these complex data.

Characterizing mutational heterogeneity in a glioblastoma patient with double recurrence

Human cancers are driven by the acquisition of somatic mutations. Separating the driving mutations from those that are random consequences of general genomic instability remains a challenge. New sequencing technology makes it possible to detect mutations that are present in only a minority of cells in a heterogeneous tumor population. We sought to leverage the power of ultra-deep sequencing to study various levels of tumor heterogeneity in the serial recurrences of a single glioblastoma multiforme patient. Our goal was to gain insight into the temporal succession of DNA base-level lesions by querying intra- and inter-tumoral cell populations in the same patient over time. We performed targeted “next-generation" sequencing on seven samples from the same patient: two foci within the primary tumor, two foci within an initial recurrence, two foci within a second recurrence, and normal blood. Our study reveals multiple levels of mutational heterogeneity. We found variable frequencies of specific EGFR, PIK3CA, PTEN, and TP53 base substitutions within individual tumor regions and across distinct regions within the same tumor. In addition, specific mutations emerge and disappear along the temporal spectrum from tumor at the time of diagnosis to second recurrence, demonstrating evolution during tumor progression. Our results shed light on the spatial and temporal complexity of brain tumors. As sequencing costs continue to decline and deep sequencing technology eventually moves into the clinic, this approach may provide guidance for treatment choices as we embark on the path to personalized cancer medicine.

An analysis of substitution, deletion and insertion mutations in cancer genes

Cancer-associated mutations in cancer genes constitute a diverse set of mutations associated with the disease. To gain insight into features of the set, substitution, deletion and insertion mutations were analysed at the nucleotide level, from the COSMIC database. The most frequent substitutions were c→t, g→a, g→t, and the most frequent codon changes were to termination codons. Deletions more than insertions, FS (frameshift) indels more than I-F (in-frame) ones, and single-nucleotide indels, were frequent. FS indels cause loss of significant fractions of proteins. The 5′-cut in FS deletions, and 5′-ligation in FS insertions, often occur between pairs of identical bases. Interestingly, the cut-site and 3′-ligation in insertions, and 3′-cut and join-pair in deletions, were each found to be the same significantly often (p < 0.001). It is suggested that these features aid the incorporation of indel mutations. Tumor suppressors undergo larger numbers of mutations, especially disruptive ones, over the entire protein length, to inactivate two alleles. Proto-oncogenes undergo fewer, less-disruptive mutations, in selected protein regions, to activate a single allele. Finally, catalogues, in ranked order, of genes mutated in each cancer, and cancers in which each gene is mutated, were created. The study highlights the nucleotide level preferences and disruptive nature of cancer mutations.

Delving into somatic variation in sporadic melanoma

Melanoma, the most aggressive form of skin cancer, has increased in incidence more rapidly than any other cancer. The completion of the human genome project and advancements in genomics technologies has allowed us to investigate genetic alterations of melanoma at a scale and depth that is unprecedented. Here, we survey the history of the different approaches taken to understand the genomics of melanoma - from early candidate genes, to gene families, to genome-wide studies. The new era of whole-exome and whole-genome sequencing has paved the way for an in-depth understanding of melanoma biology, identification of new therapeutic targets, and development of novel personalized therapies for melanoma.

Multimodel assessment of BRCA1 mutations in Taiwanese (ethnic Chinese) women with early-onset, bilateral or familial breast cancer

Although evidence suggests an importance of genetic factors in the development of breast cancer in Taiwanese (ethnic Chinese) women, including a high incidence of early-onset and secondary contralateral breast cancer, a major breast cancer predisposition gene, BRCA1, has not been well studied in this population. In fact, the carcinogenic impacts of many genetic variants of BRCA1 are unknown and classified as variants of uncertain significance (VUS). It is therefore important to establish a method to characterize the BRCA1 VUSs and understand their role in Taiwanese breast cancer patients. Accordingly, we developed a multimodel assessment strategy consisting of a prescreening portion and a validated functional assay to study breast cancer patients with early-onset, bilateral or familial breast cancer. We found germ-line BRCA1 mutations in 11.1% of our cohort and identified one novel missense mutation, c.5191C>A. Two genetic variants were initially classified as VUSs (c.1155C>T and c.5191C>A). c.1155C>T is not predicted to be deleterious in the prescreening portion of our assessment strategy. c.5191C>A, on the other hand, causes p.T1691K, which is predicted to have high deleterious probability because of significant structural alteration, a high deleterious score in the predictive programs and, clinically, triple negative characteristics in breast tumors. This mutant is confirmed by transcription activation and yeast growth-inhibition assays. In conclusion, we show as high a prevalence of germ-line BRCA1 mutation in high-risk Taiwanese patients as in Caucasians and demonstrate a useful strategy for studying BRCA1 VUSs.

Map2k4 functions as a tumor suppressor in lung adenocarcinoma and inhibits tumor cell invasion by decreasing peroxisome proliferator-activated receptor γ2 expression

MAP2K4 encodes a dual-specificity kinase (mitogen-activated protein kinase kinase 4, or MKK4) that is mutated in a variety of human malignancies, but the biochemical properties of the mutant kinases and their roles in tumorigenesis have not been fully elucidated. Here we showed that 8 out of 11 cancer-associated MAP2K4 mutations reduce MKK4 protein stability or impair its kinase activity. On the basis of findings from bioinformatic studies on human cancer cell lines with homozygous MAP2K4 loss, we posited that MKK4 functions as a tumor suppressor in lung adenocarcinomas that develop in mice owing to expression of mutant Kras and Tp53. Conditional Map2k4 inactivation in the bronchial epithelium of mice had no discernible effect alone but increased the multiplicity and accelerated the growth of incipient lung neoplasias induced by oncogenic Kras. MKK4 suppressed the invasion and metastasis of Kras-Tp53-mutant lung adenocarcinoma cells. MKK4 deficiency increased peroxisomal proliferator-activated receptor γ2 (PPARγ2) expression through noncanonical MKK4 substrates, and PPARγ2 enhanced tumor cell invasion. We conclude that Map2k4 functions as a tumor suppressor in lung adenocarcinoma and inhibits tumor cell invasion by decreasing PPARγ2 levels.

Phylomedicine: an evolutionary telescope to explore and diagnose the universe of disease mutations

Modern technologies have made the sequencing of personal genomes routine. They have revealed thousands of nonsynonymous (amino acid altering) single nucleotide variants (nSNVs) of protein-coding DNA per genome. What do these variants foretell about an individual's predisposition to diseases? The experimental technologies required to carry out such evaluations at a genomic scale are not yet available. Fortunately, the process of natural selection has lent us an almost infinite set of tests in nature. During long-term evolution, new mutations and existing variations have been evaluated for their biological consequences in countless species, and outcomes are readily revealed by multispecies genome comparisons. We review studies that have investigated evolutionary characteristics and in silico functional diagnoses of nSNVs found in thousands of disease-associated genes. We conclude that the patterns of long-term evolutionary conservation and permissible sequence divergence are essential and instructive modalities for functional assessment of human genetic variations.

Structural and functional impact of cancer-related missense somatic mutations

A number of large-scale cancer somatic genome sequencing projects are now identifying genetic alterations in cancers. Evaluation of the effects of these mutations is essential for understanding their contribution to tumorigenesis. We have used SNPs3D, a software suite originally developed for analyzing nonsynonymous germ-line variants, to identify single-base mutations with a high impact on protein structure and function. Two machine learning methods are used: one identifying mutations that destabilize protein three-dimensional structure and the other utilizing sequence conservation and detecting all types of effects on in vivo protein function. Incorporation of detailed structure information into the analysis allows detailed interpretation of the functional effects of mutations in specific cases.  Data from a set of breast and colorectal tumors were analyzed. In known cancer genes, mutations approaching 100% of mutations are found to impact protein function, supporting the view that these methods are appropriate for identifying driver mutations. Overall, 50-60% of all somatic missense mutations are predicted to have a high impact on structural stability or to more generally affect the function of the corresponding proteins. This value is similar to the fraction of all possible missense mutations that have a high impact and is much higher than the corresponding one for human population single-nucleotide polymorphisms, at about 30%. The majority of mutations in tumor suppressors destabilize protein structure, while mutations in oncogenes operate in more varied ways, including destabilization of less active conformational states. The set of high-impact mutations encompasses the possible drivers.

A new disease-specific machine learning approach for the prediction of cancer-causing missense variants

High-throughput genotyping and sequencing techniques are rapidly and inexpensively providing large amounts of human genetic variation data. Single Nucleotide Polymorphisms (SNPs) are an important source of human genome variability and have been implicated in several human diseases, including cancer. Amino acid mutations resulting from non-synonymous SNPs in coding regions may generate protein functional changes that affect cell proliferation. In this study, we developed a machine learning approach to predict cancer-causing missense variants. We present a Support Vector Machine (SVM) classifier trained on a set of 3163 cancer-causing variants and an equal number of neutral polymorphisms. The method achieve 93% overall accuracy, a correlation coefficient of 0.86, and area under ROC curve of 0.98. When compared with other previously developed algorithms such as SIFT and CHASM our method results in higher prediction accuracy and correlation coefficient in identifying cancer-causing variants.

Improving the prediction of disease-related variants using protein three-dimensional structure

Background

Single Nucleotide Polymorphisms (SNPs) are an important source of human genome variability. Non-synonymous SNPs occurring in coding regions result in single amino acid polymorphisms (SAPs) that may affect protein function and lead to pathology. Several methods attempt to estimate the impact of SAPs using different sources of information. Although sequence-based predictors have shown good performance, the quality of these predictions can be further improved by introducing new features derived from three-dimensional protein structures.

Results

In this paper, we present a structure-based machine learning approach for predicting disease-related SAPs. We have trained a Support Vector Machine (SVM) on a set of 3,342 disease-related mutations and 1,644 neutral polymorphisms from 784 protein chains. We use SVM input features derived from the protein’s sequence, structure, and function. After dataset balancing, the structure-based method (SVM-3D) reaches an overall accuracy of 85%, a correlation coefficient of 0.70, and an area under the receiving operating characteristic curve (AUC) of 0.92. When compared with a similar sequence-based predictor, SVM-3D results in an increase of the overall accuracy and AUC by 3%, and correlation coefficient by 0.06. The robustness of this improvement has been tested on different datasets and in all the cases SVM-3D performs better than previously developed methods even when compared with PolyPhen2, which explicitly considers in input protein structure information.

Conclusion

This work demonstrates that structural information can increase the accuracy of disease-related SAPs identification. Our results also quantify the magnitude of improvement on a large dataset. This improvement is in agreement with previously observed results, where structure information enhanced the prediction of protein stability changes upon mutation. Although the structural information contained in the Protein Data Bank is limiting the application and the performance of our structure-based method, we expect that SVM-3D will result in higher accuracy when more structural date become available.

Predicting the functional impact of protein mutations: application to cancer genomics

As large-scale re-sequencing of genomes reveals many protein mutations, especially in human cancer tissues, prediction of their likely functional impact becomes important practical goal. Here, we introduce a new functional impact score (FIS) for amino acid residue changes using evolutionary conservation patterns. The information in these patterns is derived from aligned families and sub-families of sequence homologs within and between species using combinatorial entropy formalism. The score performs well on a large set of human protein mutations in separating disease-associated variants (∼19 200), assumed to be strongly functional, from common polymorphisms (∼35 600), assumed to be weakly functional (area under the receiver operating characteristic curve of ∼0.86). In cancer, using recurrence, multiplicity and annotation for ∼10 000 mutations in the COSMIC database, the method does well in assigning higher scores to more likely functional mutations ('drivers'). To guide experimental prioritization, we report a list of about 1000 top human cancer genes frequently mutated in one or more cancer types ranked by likely functional impact; and, an additional 1000 candidate cancer genes with rare but likely functional mutations. In addition, we estimate that at least 5% of cancer-relevant mutations involve switch of function, rather than simply loss or gain of function.

The structural impact of cancer-associated missense mutations in oncogenes and tumor suppressors

BACKGROUND

Current large-scale cancer sequencing projects have identified large numbers of somatic mutations covering an increasing number of different cancer tissues and patients. However, the characterization of these mutations at the structural and functional level remains a challenge.

RESULTS

We present results from an analysis of the structural impact of frequent missense cancer mutations using an automated method. We find that inactivation of tumor suppressors in cancer correlates frequently with destabilizing mutations preferably in the core of the protein, while enhanced activity of oncogenes is often linked to specific mutations at functional sites. Furthermore, our results show that this alteration of oncogenic activity is often associated with mutations at ATP or GTP binding sites.

CONCLUSIONS

With our findings we can confirm and statistically validate the hypotheses for the gain-of-function and loss-of-function mechanisms of oncogenes and tumor suppressors, respectively. We show that the distinct mutational patterns can potentially be used to pre-classify newly identified cancer-associated genes with yet unknown function.

Cancer-associated loss-of-function mutations implicate DAPK3 as a tumor-suppressing kinase

Cancer kinome sequencing studies have identified several protein kinases predicted to possess driver (i.e., causal) mutations. Using bioinformatic applications, we have pinpointed DAPK3 (ZIPK) as a novel cancer-associated kinase with functional mutations. Evaluation of nonsynonymous point mutations, discovered in DAPK3 in various tumors (T112M, D161N, and P216S), reveals that all three mutations decrease or abolish kinase activity. Furthermore, phenotypic assays indicate that the three mutations observed in cancer abrogate the function of the kinase to regulate both the cell cycle and cell survival. Coexpression of wild-type (WT) and cancer mutant kinases shows that the cancer mutants dominantly inhibit the function of the WT kinase. Reconstitution of a non-small cell lung cancer cell line that harbors an endogenous mutation in DAPK3 (P216S) with WT DAPK3 resulted in decreased cellular aggregation and increased sensitivity to chemotherapy. Our results suggest that DAPK3 is a tumor suppressor in which loss-of-function mutations promote increased cell survival, proliferation, cellular aggregation, and increased resistance to chemotherapy.

Tyrosine kinase 2 variant influences T lymphocyte polarization and multiple sclerosis susceptibility

The tyrosine kinase 2 variant rs34536443 has been established as a genetic risk factor for multiple sclerosis in a variety of populations. However, the functional effect of this variant on disease pathogenesis remains unclear. This study replicated the genetic association of tyrosine kinase 2 with multiple sclerosis in a cohort of 1366 French patients and 1802 controls. Furthermore, we assessed the functional consequences of this polymorphism on human T lymphocytes by comparing the reactivity and cytokine profile of T lymphocytes isolated from individuals expressing the protective TYK2(GC) genotype with the disease-associated TYK2(GG) genotype. Our results demonstrate that the protective C allele infers decreased tyrosine kinase 2 activity, and this reduction of activity is associated with a shift in the cytokine profile favouring the secretion of Th2 cytokines. These findings suggest that the rs34536443 variant effect on multiple sclerosis susceptibility might be mediated by deviating T lymphocyte differentiation toward a Th2 phenotype. This impact of tyrosine kinase 2 on effector differentiation is likely to be of wider importance because other autoimmune diseases also have been associated with polymorphisms within tyrosine kinase 2. The modulation of tyrosine kinase 2 activity might therefore represent a new therapeutic approach for the treatment of autoimmune diseases.

Improving the assessment of the outcome of nonsynonymous SNVs with a consensus deleteriousness score, Condel

Several large ongoing initiatives that profit from next-generation sequencing technologies have driven--and in coming years will continue to drive--the emergence of long catalogs of missense single-nucleotide variants (SNVs) in the human genome. As a consequence, researchers have developed various methods and their related computational tools to classify these missense SNVs as probably deleterious or probably neutral polymorphisms. The outputs produced by each of these computational tools are of different natures and thus difficult to compare and integrate. Taking advantage of the possible complementarity between different tools might allow more accurate classifications. Here we propose an effective approach to integrating the output of some of these tools into a unified classification; this approach is based on a weighted average of the normalized scores of the individual methods (WAS). (In this paper, the approach is illustrated for the integration of five tools.) We show that this WAS outperforms each individual method in the task of classifying missense SNVs as deleterious or neutral. Furthermore, we demonstrate that this WAS can be used not only for classification purposes (deleterious versus neutral mutation) but also as an indicator of the impact of the mutation on the functionality of the mutant protein. In other words, it may be used as a deleteriousness score of missense SNVs. Therefore, we recommend the use of this WAS as a consensus deleteriousness score of missense mutations (Condel).

Protein kinase signaling networks in cancer

Protein kinases orchestrate the activation of signaling cascades in response to extracellular and intracellular stimuli to control cell growth, proliferation, and survival. The complexity of numerous intracellular signaling pathways is highlighted by the number of kinases encoded by the human genome (539) and the plethora of phosphorylation sites identified in phosphoproteomic studies. Perturbation of these signaling networks by mutations or abnormal protein expression underlies the cause of many diseases including cancer. Recent RNAi screens and cancer genomic sequencing studies have revealed that many more kinases than anticipated contribute to tumorigenesis and are potential targets for inhibitor drug development intervention. This review will highlight recent insights into known pathways essential for tumorigenesis and discuss exciting new pathways for therapeutic intervention.

Predicting disease-associated substitution of a single amino acid by analyzing residue interactions

BACKGROUND

The rapid accumulation of data on non-synonymous single nucleotide polymorphisms (nsSNPs, also called SAPs) should allow us to further our understanding of the underlying disease-associated mechanisms. Here, we use complex networks to study the role of an amino acid in both local and global structures and determine the extent to which disease-associated and polymorphic SAPs differ in terms of their interactions to other residues.

RESULTS

We found that SAPs can be well characterized by network topological features. Mutations are probably disease-associated when they occur at a site with a high centrality value and/or high degree value in a protein structure network. We also discovered that study of the neighboring residues around a mutation site can help to determine whether the mutation is disease-related or not. We compiled a dataset from the Swiss-Prot variant pages and constructed a model to predict disease-associated SAPs based on the random forest algorithm. The values of total accuracy and MCC were 83.0% and 0.64, respectively, as determined by 5-fold cross-validation. With an independent dataset, our model achieved a total accuracy of 80.8% and MCC of 0.59, respectively.

CONCLUSIONS

The satisfactory performance suggests that network topological features can be used as quantification measures to determine the importance of a site on a protein, and this approach can complement existing methods for prediction of disease-associated SAPs. Moreover, the use of this method in SAP studies would help to determine the underlying linkage between SAPs and diseases through extensive investigation of mutual interactions between residues.

Inferring the functional effects of mutation through clusters of mutations in homologous proteins

Inferring functional consequences is a bottleneck in high-throughput cancer mutation discovery and genetic association studies. Most polymorphisms and germline mutations are unlikely to have functionally significant consequences. Most cancer somatic mutations do not contribute to tumorigenesis and are not under selective pressure. Identifying and understanding functionally important mutations can clarify disease biology and lead to new therapeutic and diagnostic opportunities. We investigated the extent to which protein mutations with functional consequences are enriched in clusters at conserved positions across related proteins. We found that disease-causing mutations form clusters more than random mutations or single nucleotide polymorphisms, confirming that mutation hotspots occur at the domain level. In addition to helping to identify functionally significant mutations, analysis of clustered mutations can indicate the mechanism and consequences for protein function. Our analysis focused on somatic cancer mutations suggests functional impact for many, including singleton mutations in FGFR1, FGFR3, GFI1B, PIK3CG, RALB, RAP2B, and STK11. This provides evidence and generates mechanistic hypotheses for the contribution of such mutations to cancer. The same approach can be applied to mutations suspected of involvement in other diseases. An interactive Web application for browsing mutation clusters is available at http://www.mcluster.org.

Diverse somatic mutation patterns and pathway alterations in human cancers

The systematic characterization of somatic mutations in cancer genomes is essential for understanding the disease and for developing targeted therapeutics. Here we report the identification of 2,576 somatic mutations across approximately 1,800 megabases of DNA representing 1,507 coding genes from 441 tumours comprising breast, lung, ovarian and prostate cancer types and subtypes. We found that mutation rates and the sets of mutated genes varied substantially across tumour types and subtypes. Statistical analysis identified 77 significantly mutated genes including protein kinases, G-protein-coupled receptors such as GRM8, BAI3, AGTRL1 (also called APLNR) and LPHN3, and other druggable targets. Integrated analysis of somatic mutations and copy number alterations identified another 35 significantly altered genes including GNAS, indicating an expanded role for galpha subunits in multiple cancer types. Furthermore, our experimental analyses demonstrate the functional roles of mutant GNAO1 (a Galpha subunit) and mutant MAP2K4 (a member of the JNK signalling pathway) in oncogenesis. Our study provides an overview of the mutational spectra across major human cancers and identifies several potential therapeutic targets.

Statistical method on nonrandom clustering with application to somatic mutations in cancer

Background

Human cancer is caused by the accumulation of tumor-specific mutations in oncogenes and tumor suppressors that confer a selective growth advantage to cells. As a consequence of genomic instability and high levels of proliferation, many passenger mutations that do not contribute to the cancer phenotype arise alongside mutations that drive oncogenesis. While several approaches have been developed to separate driver mutations from passengers, few approaches can specifically identify activating driver mutations in oncogenes, which are more amenable for pharmacological intervention.

Results

We propose a new statistical method for detecting activating mutations in cancer by identifying nonrandom clusters of amino acid mutations in protein sequences. A probability model is derived using order statistics assuming that the location of amino acid mutations on a protein follows a uniform distribution. Our statistical measure is the differences between pair-wise order statistics, which is equivalent to the size of an amino acid mutation cluster, and the probabilities are derived from exact and approximate distributions of the statistical measure. Using data in the Catalog of Somatic Mutations in Cancer (COSMIC) database, we have demonstrated that our method detects well-known clusters of activating mutations in KRAS, BRAF, PI3K, and β-catenin. The method can also identify new cancer targets as well as gain-of-function mutations in tumor suppressors.

Conclusions

Our proposed method is useful to discover activating driver mutations in cancer by identifying nonrandom clusters of somatic amino acid mutations in protein sequences.

Advances in translational bioinformatics: computational approaches for the hunting of disease genes

Over a 100 years ago, William Bateson provided, through his observations of the transmission of alkaptonuria in first cousin offspring, evidence of the application of Mendelian genetics to certain human traits and diseases. His work was corroborated by Archibald Garrod (Archibald AE. The incidence of alkaptonuria: a study in chemical individuality. Lancert 1902;ii:1616-20) and William Farabee (Farabee WC. Inheritance of digital malformations in man. In: Papers of the Peabody Museum of American Archaeology and Ethnology. Cambridge, Mass: Harvard University, 1905; 65-78), who recorded the familial tendencies of inheritance of malformations of human hands and feet. These were the pioneers of the hunt for disease genes that would continue through the century and result in the discovery of hundreds of genes that can be associated with different diseases. Despite many ground-breaking discoveries during the last century, we are far from having a complete understanding of the intricate network of molecular processes involved in diseases, and we are still searching for the cures for most complex diseases. In the last few years, new genome sequencing and other high-throughput experimental techniques have generated vast amounts of molecular and clinical data that contain crucial information with the potential of leading to the next major biomedical discoveries. The need to mine, visualize and integrate these data has motivated the development of several informatics approaches that can broadly be grouped in the research area of 'translational bioinformatics'. This review highlights the latest advances in the field of translational bioinformatics, focusing on the advances of computational techniques to search for and classify disease genes.

Functional annotations improve the predictive score of human disease-related mutations in proteins

Single nucleotide polymorphisms (SNPs) are the simplest and most frequent form of human DNA variation, also valuable as genetic markers of disease susceptibility. The most investigated SNPs are missense mutations resulting in residue substitutions in the protein. Here we propose SNPs&GO, an accurate method that, starting from a protein sequence, can predict whether a mutation is disease related or not by exploiting the protein functional annotation. The scoring efficiency of SNPs&GO is as high as 82%, with a Matthews correlation coefficient equal to 0.63 over a wide set of annotated nonsynonymous mutations in proteins, including 16,330 disease-related and 17,432 neutral polymorphisms. SNPs&GO collects in unique framework information derived from protein sequence, evolutionary information, and function as encoded in the Gene Ontology terms, and outperforms other available predictive methods.

Sequence and structure signatures of cancer mutation hotspots in protein kinases

Protein kinases are the most common protein domains implicated in cancer, where somatically acquired mutations are known to be functionally linked to a variety of cancers. Resequencing studies of protein kinase coding regions have emphasized the importance of sequence and structure determinants of cancer-causing kinase mutations in understanding of the mutation-dependent activation process. We have developed an integrated bioinformatics resource, which consolidated and mapped all currently available information on genetic modifications in protein kinase genes with sequence, structure and functional data. The integration of diverse data types provided a convenient framework for kinome-wide study of sequence-based and structure-based signatures of cancer mutations. The database-driven analysis has revealed a differential enrichment of SNPs categories in functional regions of the kinase domain, demonstrating that a significant number of cancer mutations could fall at structurally equivalent positions (mutational hotspots) within the catalytic core. We have also found that structurally conserved mutational hotspots can be shared by multiple kinase genes and are often enriched by cancer driver mutations with high oncogenic activity. Structural modeling and energetic analysis of the mutational hotspots have suggested a common molecular mechanism of kinase activation by cancer mutations, and have allowed to reconcile the experimental data. According to a proposed mechanism, structural effect of kinase mutations with a high oncogenic potential may manifest in a significant destabilization of the autoinhibited kinase form, which is likely to drive tumorigenesis at some level. Structure-based functional annotation and prediction of cancer mutation effects in protein kinases can facilitate an understanding of the mutation-dependent activation process and inform experimental studies exploring molecular pathology of tumorigenesis.

Discovering cancer genes by integrating network and functional properties

BACKGROUND

Identification of novel cancer-causing genes is one of the main goals in cancer research. The rapid accumulation of genome-wide protein-protein interaction (PPI) data in humans has provided a new basis for studying the topological features of cancer genes in cellular networks. It is important to integrate multiple genomic data sources, including PPI networks, protein domains and Gene Ontology (GO) annotations, to facilitate the identification of cancer genes.

METHODS

Topological features of the PPI network, as well as protein domain compositions, enrichment of gene ontology categories, sequence and evolutionary conservation features were extracted and compared between cancer genes and other genes. The predictive power of various classifiers for identification of cancer genes was evaluated by cross validation. Experimental validation of a subset of the prediction results was conducted using siRNA knockdown and viability assays in human colon cancer cell line DLD-1.

RESULTS

Cross validation demonstrated advantageous performance of classifiers based on support vector machines (SVMs) with the inclusion of the topological features from the PPI network, protein domain compositions and GO annotations. We then applied the trained SVM classifier to human genes to prioritize putative cancer genes. siRNA knock-down of several SVM predicted cancer genes displayed greatly reduced cell viability in human colon cancer cell line DLD-1.

CONCLUSION

Topological features of PPI networks, protein domain compositions and GO annotations are good predictors of cancer genes. The SVM classifier integrates multiple features and as such is useful for prioritizing candidate cancer genes for experimental validations.

Automated inference of molecular mechanisms of disease from amino acid substitutions

MOTIVATION

Advances in high-throughput genotyping and next generation sequencing have generated a vast amount of human genetic variation data. Single nucleotide substitutions within protein coding regions are of particular importance owing to their potential to give rise to amino acid substitutions that affect protein structure and function which may ultimately lead to a disease state. Over the last decade, a number of computational methods have been developed to predict whether such amino acid substitutions result in an altered phenotype. Although these methods are useful in practice, and accurate for their intended purpose, they are not well suited for providing probabilistic estimates of the underlying disease mechanism.

RESULTS

We have developed a new computational model, MutPred, that is based upon protein sequence, and which models changes of structural features and functional sites between wild-type and mutant sequences. These changes, expressed as probabilities of gain or loss of structure and function, can provide insight into the specific molecular mechanism responsible for the disease state. MutPred also builds on the established SIFT method but offers improved classification accuracy with respect to human disease mutations. Given conservative thresholds on the predicted disruption of molecular function, we propose that MutPred can generate accurate and reliable hypotheses on the molecular basis of disease for approximately 11% of known inherited disease-causing mutations. We also note that the proportion of changes of functionally relevant residues in the sets of cancer-associated somatic mutations is higher than for the inherited lesions in the Human Gene Mutation Database which are instead predicted to be characterized by disruptions of protein structure.

AVAILABILITY

http://mutdb.org/mutpred

CONTACT

predrag@indiana.edu; smooney@buckinstitute.org.

Cancer-specific high-throughput annotation of somatic mutations: computational prediction of driver missense mutations

Large-scale sequencing of cancer genomes has uncovered thousands of DNA alterations, but the functional relevance of the majority of these mutations to tumorigenesis is unknown. We have developed a computational method, called Cancer-specific High-throughput Annotation of Somatic Mutations (CHASM), to identify and prioritize those missense mutations most likely to generate functional changes that enhance tumor cell proliferation. The method has high sensitivity and specificity when discriminating between known driver missense mutations and randomly generated missense mutations (area under receiver operating characteristic curve, >0.91; area under Precision-Recall curve, >0.79). CHASM substantially outperformed previously described missense mutation function prediction methods at discriminating known oncogenic mutations in P53 and the tyrosine kinase epidermal growth factor receptor. We applied the method to 607 missense mutations found in a recent glioblastoma multiforme sequencing study. Based on a model that assumed the glioblastoma multiforme mutations are a mixture of drivers and passengers, we estimate that 8% of these mutations are drivers, causally contributing to tumorigenesis.

WebGMAP: a web service for mapping and aligning cDNA sequences to genomes

The genomes of thousands of organisms are being sequenced, often with accompanying sequences of cDNAs or ESTs. One of the great challenges in bioinformatics is to make these genomic sequences and genome annotations accessible in a user-friendly manner to general biologists to address interesting biological questions. We have created an open-access web service called WebGMAP (http://www.bioinfolab.org/software/webgmap) that seamlessly integrates cDNA-genome alignment tools, such as GMAP, with easy-to-use data visualization and mining tools. This web service is intended to facilitate community efforts in improving genome annotation, determining accurate gene structures and their variations, and exploring important biological processes such as alternative splicing and alternative polyadenylation. For routine sequence analysis, WebGMAP provides a web-based sequence viewer with many useful functions, including nucleotide positioning, six-frame translations, sequence reverse complementation, and imperfect motif detection and alignment. WebGMAP also provides users with the ability to sort, filter and search for individual cDNA sequences and cDNA-genome alignments. Our EST-Genome-Browser can display annotated gene structures and cDNA-genome alignments at scales from 100 to 50 000 nt. With its ability to highlight base differences between query cDNAs and the genome, our EST-Genome-Browser allows biologists to discover potential point or insertion-deletion variations from cDNA-genome alignments.

From cancer genomes to cancer models: bridging the gaps

Cancer genome projects are now being expanded in an attempt to provide complete landscapes of the mutations that exist in tumours. Although the importance of cataloguing genome variations is well recognized, there are obvious difficulties in bridging the gaps between high-throughput resequencing information and the molecular mechanisms of cancer evolution. Here, we describe the current status of the high-throughput genomic technologies, and the current limitations of the associated computational analysis and experimental validation of cancer genetic variants. We emphasize how the current cancer-evolution models will be influenced by the high-throughput approaches, in particular through efforts devoted to monitoring tumour progression, and how, in turn, the integration of data and models will be translated into mechanistic knowledge and clinical applications.

Computational prediction of the functional effects of amino acid substitutions in signal peptides using a model-based approach

Signal peptides are N-terminal sequences that mediate the targeting and translocation of secreted or cell-surface proteins to the endoplasmic reticulum (ER) membrane. Because of the variability among signal peptides, traditional methods for predicting the effects of an amino acid substitution based on sequence conservation methods may be limited in their use. To address this, we present a scoring function that assesses the effects of an amino acid change within the signal peptide by using data from SignalP, a signal peptide prediction algorithm. Our score incorporates the maximum alterations of the C- and S-scores from SignalP between original and changed versions of the signal peptide. We demonstrate that this metric can discriminate disease-associated mutations from single nucleotide polymorphisms (SNPs) in signal peptides. We further show that polymorphisms with low minor allele frequency (MAF) are more likely to affect the function of the signal peptide. In conjunction with Sorting Intolerant From Tolerant (SIFT), a conservation-based amino acid substitution prediction method, our approach classifies such changes to signal peptides more accurately than other known alternatives, including D-score-based methods. We also examine experimentally characterized mutations and find that our metric minimizes false positives and can predict whether the mutation will affect cleavage or translocation. Finally, we apply our approach to a set of recently produced large-scale cancer somatic mutations from colon and breast cancers and generate a prioritized list of mutations in signal peptides that might impair protein function.

Identification of an indispensable role for tyrosine kinase 2 in CTL-mediated tumor surveillance

We showed previously that Tyk2(-/-) natural killer cells lack the ability to lyse leukemic cells. As a consequence, the animals are leukemia prone. Here, we show that the impaired tumor surveillance extends to T cells. Challenging Tyk2(-/-) mice with EL4 thymoma significantly decreased disease latency. The crucial role of Tyk2 for CTL function was further characterized using the ovalbumin-expressing EG7 cells. Tyk2(-/-) OT-1 mice developed EG7-induced tumors significantly faster compared with wild-type (wt) controls. In vivo assays confirmed the defect in CD8(+) cytotoxicity on Tyk2 deficiency and clearly linked it to type I IFN signaling. An impaired CTL activity was only observed in IFNAR1(-/-) animals but not on IFNgamma or IL12p35 deficiency. Accordingly, EG7-induced tumors grew faster in IFNAR1(-/-) and Tyk2(-/-) but not in IFNgamma(-/-) or IL12p35(-/-) mice. Adoptive transfer experiments defined a key role of Tyk2 in CTL-mediated tumor surveillance. In contrast to wt OT-1 cells, Tyk2(-/-) OT-1 T cells were incapable of controlling EG7-induced tumor growth.

Cancer driver mutations in protein kinase genes

Recent studies investigating the genetic determinants of cancer suggest that some of the genetic alterations contributing to tumorigenesis may be inherited, but the vast majority is somatically acquired during the transition of a normal cell to a cancer cell. A systematic understanding of the genetic and molecular determinants of cancers has already begun to have a transformative effect on the study and treatment of cancer, particularly through the identification of a range of genetic alterations in protein kinase genes, which are highly associated with the disease. Since kinases are prominent therapeutic targets for intervention within the cancer cell, studying the impact that genomic alterations within them have on cancer initiation, progression, and treatment is both logical and timely. In fact, recent sequencing and resequencing (i.e., polymorphism identification) efforts have catalyzed the quest for protein kinase 'driver' mutations (i.e., those genetic alterations which contribute to the transformation of a normal cell to a proliferating cancerous cell) in distinction to kinase 'passenger' mutations which reflect mutations that merely build up in course of normal and unchecked (i.e., cancerous) somatic cell replication and proliferation. In this review, we discuss the recent progress in the discovery and functional characterization of protein kinase cancer driver mutations and the implications of this progress for understanding tumorigenesis as well as the design of 'personalized' cancer therapeutics that target an individual's unique mutational profile.

MoKCa database--mutations of kinases in cancer

Members of the protein kinase family are amongst the most commonly mutated genes in human cancer, and both mutated and activated protein kinases have proved to be tractable targets for the development of new anticancer therapies The MoKCa database (Mutations of Kinases in Cancer, http://strubiol.icr.ac.uk/extra/mokca) has been developed to structurally and functionally annotate, and where possible predict, the phenotypic consequences of mutations in protein kinases implicated in cancer. Somatic mutation data from tumours and tumour cell lines have been mapped onto the crystal structures of the affected protein domains. Positions of the mutated amino-acids are highlighted on a sequence-based domain pictogram, as well as a 3D-image of the protein structure, and in a molecular graphics package, integrated for interactive viewing. The data associated with each mutation is presented in the Web interface, along with expert annotation of the detailed molecular functional implications of the mutation. Proteins are linked to functional annotation resources and are annotated with structural and functional features such as domains and phosphorylation sites. MoKCa aims to provide assessments available from multiple sources and algorithms for each potential cancer-associated mutation, and present these together in a consistent and coherent fashion to facilitate authoritative annotation by cancer biologists and structural biologists, directly involved in the generation and analysis of new mutational data.

Gain and loss of phosphorylation sites in human cancer

MOTIVATION

Coding-region mutations in human genes are responsible for a diverse spectrum of diseases and phenotypes. Among lesions that have been studied extensively, there are insights into several of the biochemical functions disrupted by disease-causing mutations. Currently, there are more than 60 000 coding-region mutations associated with inherited disease catalogued in the Human Gene Mutation Database (HGMD, August 2007) and more than 70 000 polymorphic amino acid substitutions recorded in dbSNP (dbSNP, build 127). Understanding the mechanism and contribution these variants make to a clinical phenotype is a formidable problem.

RESULTS

In this study, we investigate the role of phosphorylation in somatic cancer mutations and inherited diseases. Somatic cancer mutation datasets were shown to have a significant enrichment for mutations that cause gain or loss of phosphorylation when compared to our control datasets (putatively neutral nsSNPs and random amino acid substitutions). Of the somatic cancer mutations, those in kinase genes represent the most enriched set of mutations that disrupt phosphorylation sites, suggesting phosphorylation target site mutation is an active cause of phosphorylation deregulation. Overall, this evidence suggests both gain and loss of a phosphorylation site in a target protein may be important features for predicting cancer-causing mutations and may represent a molecular cause of disease for a number of inherited and somatic mutations.

Prioritization of candidate cancer genes--an aid to oncogenomic studies

The development of techniques for oncogenomic analyses such as array comparative genomic hybridization, messenger RNA expression arrays and mutational screens have come to the fore in modern cancer research. Studies utilizing these techniques are able to highlight panels of genes that are altered in cancer. However, these candidate cancer genes must then be scrutinized to reveal whether they contribute to oncogenesis or are coincidental and non-causative. We present a computational method for the prioritization of candidate (i) proto-oncogenes and (ii) tumour suppressor genes from oncogenomic experiments. We constructed computational classifiers using different combinations of sequence and functional data including sequence conservation, protein domains and interactions, and regulatory data. We found that these classifiers are able to distinguish between known cancer genes and other human genes. Furthermore, the classifiers also discriminate candidate cancer genes from a recent mutational screen from other human genes. We provide a web-based facility through which cancer biologists may access our results and we propose computational cancer gene classification as a useful method of prioritizing candidate cancer genes identified in oncogenomic studies.

Breast cancer-associated mutations in metalloprotease disintegrin ADAM12 interfere with the intracellular trafficking and processing of the protein

ADAM12 has recently emerged as a Candidate Cancer Gene in a comprehensive genetic analysis of human breast cancers. Three somatic mutations in ADAM12 were observed at significant frequencies in breast cancers: D301H, G479E and L792F. The first 2 of these mutations involve highly conserved residues in ADAM12, and our computational sequence analysis confirms that they may be cancer-related. We show that the corresponding mutations in mouse ADAM12 inhibit the proteolytic processing and activation of ADAM12 in NIH3T3, COS-7, CHO-K1 cells and in MCF-7 breast cancer cells. The D/H and G/E ADAM12 mutants exert a dominant-negative effect on the processing of the wild-type ADAM12. Immunofluorescence analysis and cell surface biotinylation experiments demonstrate that the D/H and G/E mutants are retained inside the cell and are not transported to the cell surface. Consequently, the D/H and G/E mutants, unlike the wild-type ADAM12, are not capable of shedding Delta-like l, a ligand for Notch receptor, at the cell surface, or of stimulating cell migration. Our results suggest that the breast cancer-associated mutations interfere with the intracellular trafficking of ADAM12 and result in loss of the functional ADAM12 at the cell surface.

Prediction of cancer driver mutations in protein kinases

A large number of somatic mutations accumulate during the process of tumorigenesis. A subset of these mutations contribute to tumor progression (known as "driver" mutations) whereas the majority of these mutations are effectively neutral (known as "passenger" mutations). The ability to differentiate between drivers and passengers will be critical to the success of upcoming large-scale cancer DNA resequencing projects. Here we show a method capable of discriminating between drivers and passengers in the most frequently cancer-associated protein family, protein kinases. We apply this method to multiple cancer data sets, validating its accuracy by showing that it is capable of identifying known drivers, has excellent agreement with previous statistical estimates of the frequency of drivers, and provides strong evidence that predicted drivers are under positive selection by various sequence and structural analyses. Furthermore, we identify particular positions in protein kinases that seem to play a role in oncogenesis. Finally, we provide a ranked list of candidate driver mutations.

Transcriptome sequencing of malignant pleural mesothelioma tumors

Cancers arise by the gradual accumulation of mutations in multiple genes. We now use shotgun pyrosequencing to characterize RNA mutations and expression levels unique to malignant pleural mesotheliomas (MPMs) and not present in control tissues. On average, 266 Mb of cDNA were sequenced from each of four MPMs, from a control pulmonary adenocarcinoma (ADCA), and from normal lung tissue. Previously observed differences in MPM RNA expression levels were confirmed. Point mutations were identified by using criteria that require the presence of the mutation in at least four reads and in both cDNA strands and the absence of the mutation from sequence databases, normal adjacent tissues, and other controls. In the four MPMs, 15 nonsynonymous mutations were discovered: 7 were point mutations, 3 were deletions, 4 were exclusively expressed as a consequence of imputed epigenetic silencing, and 1 was putatively expressed as a consequence of RNA editing. Notably, each MPM had a different mutation profile, and no mutated gene was previously implicated in MPM. Of the seven point mutations, three were observed in at least one tumor from 49 other MPM patients. The mutations were in genes that could be causally related to cancer and included XRCC6, PDZK1IP1, ACTR1A, and AVEN.

Somatic mutations and germline sequence variants in the expressed tyrosine kinase genes of patients with de novo acute myeloid leukemia

Activating mutations in tyrosine kinase (TK) genes (eg, FLT3 and KIT) are found in more than 30% of patients with de novo acute myeloid leukemia (AML); many groups have speculated that mutations in other TK genes may be present in the remaining 70%. We performed high-throughput resequencing of the kinase domains of 26 TK genes (11 receptor TK; 15 cytoplasmic TK) expressed in most AML patients using genomic DNA from the bone marrow (tumor) and matched skin biopsy samples ("germline") from 94 patients with de novo AML; sequence variants were validated in an additional 94 AML tumor samples (14.3 million base pairs of sequence were obtained and analyzed). We identified known somatic mutations in FLT3, KIT, and JAK2 TK genes at the expected frequencies and found 4 novel somatic mutations, JAK1(V623A), JAK1(T478S), DDR1(A803V), and NTRK1(S677N), once each in 4 respective patients of 188 tested. We also identified novel germline sequence changes encoding amino acid substitutions (ie, nonsynonymous changes) in 14 TK genes, including TYK2, which had the largest number of nonsynonymous sequence variants (11 total detected). Additional studies will be required to define the roles that these somatic and germline TK gene variants play in AML pathogenesis.

Phase II study of efficacy and safety of bevacizumab in combination with chemotherapy or erlotinib compared with chemotherapy alone for treatment of recurrent or refractory non small-cell lung cancer

PURPOSE

Bevacizumab, a humanized anti-vascular endothelial growth factor monoclonal antibody, and erlotinib, a reversible, orally available epidermal growth factor receptor tyrosine kinase inhibitor, have demonstrated evidence of a survival benefit in the treatment of non-small-cell lung cancer (NSCLC). A single-arm phase I and II study of bevacizumab plus erlotinib demonstrated encouraging efficacy, with a favorable safety profile.

PATIENTS AND METHODS

A multicenter, randomized phase II trial evaluated the safety of combining bevacizumab with either chemotherapy (docetaxel or pemetrexed) or erlotinib and preliminarily assessed these combinations versus chemotherapy alone, as measured by progression-free survival (PFS). All patients had histologically confirmed nonsquamous NSCLC that had progressed during or after one platinum-based regimen.

RESULTS

One hundred twenty patients were randomly assigned and treated. No unexpected adverse events were noted. Fewer patients (13%) in the bevacizumab-erlotinib arm discontinued treatment as a result of adverse events than in the chemotherapy alone (24%) or bevacizumab-chemotherapy (28%) arms. The incidence of grade 5 hemorrhage in patients receiving bevacizumab was 5.1%. Although not statistically significant, relative to chemotherapy alone, the risk of disease progression or death was 0.66 (95% CI, 0.38 to 1.16) among patients treated with bevacizumab-chemotherapy and 0.72 (95% CI, 0.42 to 1.23) among patients treated with bevacizumab-erlotinib. One-year survival rate was 57.4% for bevacizumab-erlotinib and 53.8% for bevacizumab-chemotherapy compared with 33.1% for chemotherapy alone.

CONCLUSION

Results for PFS and overall survival favor combination of bevacizumab with either chemotherapy or erlotinib over chemotherapy alone in the second-line setting. No unexpected safety signals were noted. The rate of fatal pulmonary hemorrhage was consistent with previous bevacizumab trials. The toxicity profile of the bevacizumab-erlotinib combination is favorable compared with either chemotherapy-containing group.

CanPredict: a computational tool for predicting cancer-associated missense mutations

Various cancer genome projects are underway to identify novel mutations that drive tumorigenesis. While these screens will generate large data sets, the majority of identified missense changes are likely to be innocuous passenger mutations or polymorphisms. As a result, it has become increasingly important to develop computational methods for distinguishing functionally relevant mutations from other variations. We previously developed an algorithm, and now present the web application, CanPredict (http://www.canpredict.org/ or http://www.cgl.ucsf.edu/Research/genentech/canpredict/), to allow users to determine if particular changes are likely to be cancer-associated. The impact of each change is measured using two known methods: Sorting Intolerant From Tolerant (SIFT) and the Pfam-based LogR.E-value metric. A third method, the Gene Ontology Similarity Score (GOSS), provides an indication of how closely the gene in which the variant resides resembles other known cancer-causing genes. Scores from these three algorithms are analyzed by a random forest classifier which then predicts whether a change is likely to be cancer-associated. CanPredict fills an important need in cancer biology and will enable a large audience of biologists to determine which mutations are the most relevant for further study.