Can tandem alternative splicing and evasion of premature termination codon surveillance contribute to attenuated Peutz-Jeghers syndrome?

Alternative use of short distance tandem sites such as NAGNn AG are a common mechanism of alternative splicing; however, single nucleotide variants are rarely reported as likely to generate or to disrupt tandem splice sites. We identify a pathogenic intron 5 STK11 variant (NM_000455.4:c.[735-6A>G];[=]) segregating with the mucocutaneous features but not the hamartomatous polyps of Peutz-Jeghers syndrome in two individuals. By RNAseq analysis of peripheral blood mRNA, this variant was shown to generate a novel and preferentially used tandem proximal splice acceptor (AAGTGAAG). The variant transcript (NM_000455.4:c.734_734 + 1insTGAAG), which encodes a frameshift (p.[Tyr246Glufs*43]) constituted 36%-43% of STK11 transcripts suggesting partial escape from nonsense mediated mRNA decay and translation of a truncated protein. A review of the ClinVar database identified other similar variants. We suggest that nucleotide changes creating or disrupting tandem alternative splice sites are a pertinent disease mechanism and require contextualization for clinical reporting. Additionally, we hypothesize that some pathogenic STK11 variants cause an attenuated phenotype.

Pan-cancer RNA-seq data stratifies tumours by some hallmarks of cancer.

Numerous genetic and epigenetic alterations cause functional changes in cell biology underlying cancer. These hallmark functional changes constitute potentially tissue-independent anticancer therapeutic targets. We hypothesized that RNA-Seq identifies gene expression changes that underly those hallmarks, and thereby defines relevant therapeutic targets. To test this hypothesis, we analysed the publicly available TCGA-TARGET-GTEx gene expression data set from the University of California Santa CruzToil recompute project using WGCNA to delineate co-correlated 'modules' from tumour gene expression profiles and functional enrichment of these modules to hierarchically cluster tumours. This stratified tumours according to T cell activation, NK-cell activation, complement cascade, ATM, Rb, angiogenic, MAPK, ECM receptor and histone modification signalling. These correspond to the cancer hallmarks of avoiding immune destruction, tumour-promoting inflammation, evading growth suppressors, inducing angiogenesis, sustained proliferative signalling, activating invasion and metastasis, and genome instability and mutation. This approach did not detect pathways corresponding to the cancer enabling replicative immortality, resisting cell death or deregulating cellular energetics hallmarks. We conclude that RNA-Seq stratifies tumours along some, but not all, hallmarks of cancer and, therefore, could be used in conjunction with other analyses collectively to inform precision therapy.

Gene-based sequencing identifies lipid-influencing variants with ethnicity-specific effects in African Americans.

Although a considerable proportion of serum lipids loci identified in European ancestry individuals (EA) replicate in African Americans (AA), interethnic differences in the distribution of serum lipids suggest that some genetic determinants differ by ethnicity. We conducted a comprehensive evaluation of five lipid candidate genes to identify variants with ethnicity-specific effects. We sequenced ABCA1, LCAT, LPL, PON1, and SERPINE1 in 48 AA individuals with extreme serum lipid concentrations (high HDLC/low TG or low HDLC/high TG). Identified variants were genotyped in the full population-based sample of AA (n = 1694) and tested for an association with serum lipids. rs328 (LPL) and correlated variants were associated with higher HDLC and lower TG. Interestingly, a stronger effect was observed on a "European" vs. "African" genetic background at this locus. To investigate this effect, we evaluated the region among West Africans (WA). For TG, the effect size among WA was the same in AA with only African local ancestry (2-3% lower TG), while the larger association among AA with local European ancestry matched previous reports in EA (10%). For HDLC, there was no association with rs328 in AA with only African local ancestry or in WA, while the association among AA with European local ancestry was much greater than what has been observed for EA (15 vs. ∼ 5 mg/dl), suggesting an interaction with an environmental or genetic factor that differs by ethnicity. Beyond this ancestry effect, the importance of African ancestry-focused, sequence-based work was also highlighted by serum lipid associations of variants that were in higher frequency (or present only) among those of African ancestry. By beginning our study with the sequence variation present in AA individuals, investigating local ancestry effects, and seeking replication in WA, we were able to comprehensively evaluate the role of a set of candidate genes in serum lipids in AA.

Replicate exome-sequencing in a multiple-generation family: improved interpretation of next-generation sequencing data.

BACKGROUND: Whole-exome sequencing (WES) is rapidly evolving into a tool of choice for rapid, and inexpensive identification of molecular genetic lesions within targeted regions of the human genome. While biases in WES coverage of nucleotides in targeted regions are recognized, it is not well understood how repetition of WES improves the interpretation of sequencing results in a clinical diagnostic setting. METHOD: To address this, we compared independently generated exome-capture of six individuals from three-generations sequenced in triplicate. This generated between 48x-86x mean target depth of high-quality mapped bases (>Q20) for each technical replicate library. Cumulatively, we achieved 179 - 208x average target coverage for each individual in the pedigree. Using this experimental design, we evaluated stochastics in WES interpretation, genotyping sensitivity, and accuracy to detect de novo variants. RESULTS: In this study, we show that repetition of WES improved the interpretation of the capture target regions after aggregating the data (93.5 - 93.9 %). Compared to 81.2 - 89.6 % (50.2-55.4 Mb of 61.7 M) coverage of targeted bases at ≥20x in the individual technical replicates, the aggregated data covered 93.5 - 93.9 % of targeted bases (57.7 - 58.0 of 61.7 M) at ≥20x threshold, suggesting a 4.3 - 12.7 % improvement in coverage. Each individual's aggregate dataset recovered 3.4 - 6.4 million bases within variable targeted regions. We uncovered technical variability (2-5 %) inherent to WES technique. We also show improved interpretation in assessing clinically important regions that lack interpretation under current conditions, affecting 12-16 of the 56 genes recommended for secondary analysis by American College of Medical Genetics (ACMG). We demonstrate that comparing technical replicate WES datasets and their derived aggregate data can effectively address overall WES genotyping discrepancies. CONCLUSION: We describe a method to evaluate the reproducibility and stochastics in exome library preparation, and delineate the advantages of aggregating the data derived from technical replicates. The implications of this study are directly applicable to improved experimental design and provide an opportunity to rapidly, efficiently, and accurately arrive at reliable candidate nucleotide variants.

Circadian changes in long noncoding RNAs in the pineal gland.

Long noncoding RNAs (lncRNAs) play a broad range of biological roles, including regulation of expression of genes and chromosomes. Here, we present evidence that lncRNAs are involved in vertebrate circadian biology. Differential night/day expression of 112 lncRNAs (0.3 to >50 kb) occurs in the rat pineal gland, which is the source of melatonin, the hormone of the night. Approximately one-half of these changes reflect nocturnal increases. Studies of eight lncRNAs with 2- to >100-fold daily rhythms indicate that, in most cases, the change results from neural stimulation from the central circadian oscillator in the suprachiasmatic nucleus (doubling time = 0.5-1.3 h). Light exposure at night rapidly reverses (halving time = 9-32 min) levels of some of these lncRNAs. Organ culture studies indicate that expression of these lncRNAs is regulated by norepinephrine acting through cAMP. These findings point to a dynamic role of lncRNAs in the circadian system.

Identification of candidate genes involved in coronary artery calcification by transcriptome sequencing of cell lines.

BACKGROUND: Massively-parallel cDNA sequencing (RNA-Seq) is a new technique that holds great promise for cardiovascular genomics. Here, we used RNA-Seq to study the transcriptomes of matched coronary artery disease cases and controls in the ClinSeq® study, using cell lines as tissue surrogates. RESULTS: Lymphoblastoid cell lines (LCLs) from 16 cases and controls representing phenotypic extremes for coronary calcification were cultured and analyzed using RNA-Seq. All cell lines were then independently re-cultured and along with another set of 16 independent cases and controls, were profiled with Affymetrix microarrays to perform a technical validation of the RNA-Seq results. Statistically significant changes (p < 0.05) were detected in 186 transcripts, many of which are expressed at extremely low levels (5-10 copies/cell), which we confirmed through a separate spike-in control RNA-Seq experiment. Next, by fitting a linear model to exon-level RNA-Seq read counts, we detected signals of alternative splicing in 18 transcripts. Finally, we used the RNA-Seq data to identify differential expression (p < 0.0001) in eight previously unannotated regions that may represent novel transcripts. Overall, differentially expressed genes showed strong enrichment (p = 0.0002) for prior association with cardiovascular disease. At the network level, we found evidence for perturbation in pathways involving both cardiovascular system development and function as well as lipid metabolism. CONCLUSIONS: We present a pilot study for transcriptome involvement in coronary artery calcification and demonstrate how RNA-Seq analyses using LCLs as a tissue surrogate may yield fruitful results in a clinical sequencing project. In addition to canonical gene expression, we present candidate variants from alternative splicing and novel transcript detection, which have been unexplored in the context of this disease.

Reproduction and immunity-driven natural selection in the human WFDC locus.

The whey acidic protein (WAP) four-disulfide core domain (WFDC) locus located on human chromosome 20q13 spans 19 genes with WAP and/or Kunitz domains. These genes participate in antimicrobial, immune, and tissue homoeostasis activities. Neighboring SEMG genes encode seminal proteins Semenogelin 1 and 2 (SEMG1 and SEMG2). WFDC and SEMG genes have a strikingly high rate of amino acid replacement (dN/dS), indicative of responses to adaptive pressures during vertebrate evolution. To better understand the selection pressures acting on WFDC genes in human populations, we resequenced 18 genes and 54 noncoding segments in 71 European (CEU), African (YRI), and Asian (CHB + JPT) individuals. Overall, we identified 484 single-nucleotide polymorphisms (SNPs), including 65 coding variants (of which 49 are nonsynonymous differences). Using classic neutrality tests, we confirmed the signature of short-term balancing selection on WFDC8 in Europeans and a signature of positive selection spanning genes PI3, SEMG1, SEMG2, and SLPI. Associated with the latter signal, we identified an unusually homogeneous-derived 100-kb haplotype with a frequency of 88% in Asian populations. A putative candidate variant targeted by selection is Thr56Ser in SEMG1, which may alter the proteolytic profile of SEMG1 and antimicrobial activities of semen. All the well-characterized genes residing in the WDFC locus encode proteins that appear to have a role in immunity and/or fertility, two processes that are often associated with adaptive evolution. This study provides further evidence that the WFDC and SEMG loci have been under strong adaptive pressure within the short timescale of modern humans.

Whole-exome sequencing identifies homozygous AFG3L2 mutations in a spastic ataxia-neuropathy syndrome linked to mitochondrial m-AAA proteases.

We report an early onset spastic ataxia-neuropathy syndrome in two brothers of a consanguineous family characterized clinically by lower extremity spasticity, peripheral neuropathy, ptosis, oculomotor apraxia, dystonia, cerebellar atrophy, and progressive myoclonic epilepsy. Whole-exome sequencing identified a homozygous missense mutation (c.1847G>A; p.Y616C) in AFG3L2, encoding a subunit of an m-AAA protease. m-AAA proteases reside in the mitochondrial inner membrane and are responsible for removal of damaged or misfolded proteins and proteolytic activation of essential mitochondrial proteins. AFG3L2 forms either a homo-oligomeric isoenzyme or a hetero-oligomeric complex with paraplegin, a homologous protein mutated in hereditary spastic paraplegia type 7 (SPG7). Heterozygous loss-of-function mutations in AFG3L2 cause autosomal-dominant spinocerebellar ataxia type 28 (SCA28), a disorder whose phenotype is strikingly different from that of our patients. As defined in yeast complementation assays, the AFG3L2(Y616C) gene product is a hypomorphic variant that exhibited oligomerization defects in yeast as well as in patient fibroblasts. Specifically, the formation of AFG3L2(Y616C) complexes was impaired, both with itself and to a greater extent with paraplegin. This produced an early-onset clinical syndrome that combines the severe phenotypes of SPG7 and SCA28, in additional to other "mitochondrial" features such as oculomotor apraxia, extrapyramidal dysfunction, and myoclonic epilepsy. These findings expand the phenotype associated with AFG3L2 mutations and suggest that AFG3L2-related disease should be considered in the differential diagnosis of spastic ataxias.

Predisposition to cancer caused by genetic and functional defects of mammalian Atad5.

ATAD5, the human ortholog of yeast Elg1, plays a role in PCNA deubiquitination. Since PCNA modification is important to regulate DNA damage bypass, ATAD5 may be important for suppression of genomic instability in mammals in vivo. To test this hypothesis, we generated heterozygous (Atad5(+/m)) mice that were haploinsuffficient for Atad5. Atad5(+/m) mice displayed high levels of genomic instability in vivo, and Atad5(+/m) mouse embryonic fibroblasts (MEFs) exhibited molecular defects in PCNA deubiquitination in response to DNA damage, as well as DNA damage hypersensitivity and high levels of genomic instability, apoptosis, and aneuploidy. Importantly, 90% of haploinsufficient Atad5(+/m) mice developed tumors, including sarcomas, carcinomas, and adenocarcinomas, between 11 and 20 months of age. High levels of genomic alterations were evident in tumors that arose in the Atad5(+/m) mice. Consistent with a role for Atad5 in suppressing tumorigenesis, we also identified somatic mutations of ATAD5 in 4.6% of sporadic human endometrial tumors, including two nonsense mutations that resulted in loss of proper ATAD5 function. Taken together, our findings indicate that loss-of-function mutations in mammalian Atad5 are sufficient to cause genomic instability and tumorigenesis.

Massively parallel sequencing of exons on the X chromosome identifies RBM10 as the gene that causes a syndromic form of cleft palate.

Micrognathia, glossoptosis, and cleft palate comprise one of the most common malformation sequences, Robin sequence. It is a component of the TARP syndrome, talipes equinovarus, atrial septal defect, Robin sequence, and persistent left superior vena cava. This disorder is X-linked and severe, with apparently 100% pre- or postnatal lethality in affected males. Here we characterize a second family with TARP syndrome, confirm linkage to Xp11.23-q13.3, perform massively parallel sequencing of X chromosome exons, filter the results via a number of criteria including the linkage region, use a unique algorithm to characterize sequence changes, and show that TARP syndrome is caused by mutations in the RBM10 gene, which encodes RNA binding motif 10. We further show that this previously uncharacterized gene is expressed in midgestation mouse embryos in the branchial arches and limbs, consistent with the human phenotype. We conclude that massively parallel sequencing is useful to characterize large candidate linkage intervals and that it can be used successfully to allow identification of disease-causing gene mutations.

Compound heterozygosity for loss-of-function lysyl-tRNA synthetase mutations in a patient with peripheral neuropathy.

Charcot-Marie-Tooth (CMT) disease comprises a genetically and clinically heterogeneous group of peripheral nerve disorders characterized by impaired distal motor and sensory function. Mutations in three genes encoding aminoacyl-tRNA synthetases (ARSs) have been implicated in CMT disease primarily associated with an axonal pathology. ARSs are ubiquitously expressed, essential enzymes responsible for charging tRNA molecules with their cognate amino acids. To further explore the role of ARSs in CMT disease, we performed a large-scale mutation screen of the 37 human ARS genes in a cohort of 355 patients with a phenotype consistent with CMT. Here we describe three variants (p.Leu133His, p.Tyr173SerfsX7, and p.Ile302Met) in the lysyl-tRNA synthetase (KARS) gene in two patients from this cohort. Functional analyses revealed that two of these mutations (p.Leu133His and p.Tyr173SerfsX7) severely affect enzyme activity. Interestingly, both functional variants were found in a single patient with CMT disease and additional neurological and non-neurological sequelae. Based on these data, KARS becomes the fourth ARS gene associated with CMT disease, indicating that this family of enzymes is specifically critical for axon function.

Genome-wide ChIP-Seq reveals a dramatic shift in the binding of the transcription factor erythroid Kruppel-like factor during erythrocyte differentiation.

Erythropoiesis is dependent on the activity of transcription factors, including the erythroid-specific erythroid Kruppel-like factor (EKLF). ChIP followed by massively parallel sequencing (ChIP-Seq) is a powerful, unbiased method to map trans-factor occupancy. We used ChIP-Seq to study the interactome of EKLF in mouse erythroid progenitor cells and more differentiated erythroblasts. We correlated these results with the nuclear distribution of EKLF, RNA-Seq analysis of the transcriptome, and the occupancy of other erythroid transcription factors. In progenitor cells, EKLF is found predominantly at the periphery of the nucleus, where EKLF primarily occupies the promoter regions of genes and acts as a transcriptional activator. In erythroblasts, EKLF is distributed throughout the nucleus, and erythroblast-specific EKLF occupancy is predominantly in intragenic regions. In progenitor cells, EKLF modulates general cell growth and cell cycle regulatory pathways, whereas in erythroblasts EKLF is associated with repression of these pathways. The EKLF interactome shows very little overlap with the interactomes of GATA1, GATA2, or TAL1, leading to a model in which EKLF directs programs that are independent of those regulated by the GATA factors or TAL1.

An analysis of exome sequencing for diagnostic testing of the genes associated with muscle disease and spastic paraplegia.

In this study, we assess exome sequencing (ES) as a diagnostic alternative for genetically heterogeneous disorders. Because ES readily identified a previously reported homozygous mutation in the CAPN3 gene for an individual with an undiagnosed limb girdle muscular dystrophy, we evaluated ES as a generalizable clinical diagnostic tool by assessing the targeting efficiency and sequencing coverage of 88 genes associated with muscle disease (MD) and spastic paraplegia (SPG). We used three exome-capture kits on 125 individuals. Exons constituting each gene were defined using the UCSC and CCDS databases. The three exome-capture kits targeted 47-92% of bases within the UCSC-defined exons and 97-99% of bases within the CCDS-defined exons. An average of 61.2-99.5% and 19.1-99.5% of targeted bases per gene were sequenced to 20X coverage within the CCDS-defined MD and SPG coding exons, respectively. Greater than 95-99% of targeted known mutation positions were sequenced to ≥1X coverage and 55-87% to ≥20X coverage in every exome. We conclude, therefore, that ES is a rapid and efficient first-tier method to screen for mutations, particularly within the CCDS annotated exons, although its application requires disclosure of the extent of coverage for each targeted gene and supplementation with second-tier Sanger sequencing for full coverage.

The ClinSeq Project: piloting large-scale genome sequencing for research in genomic medicine.

ClinSeq is a pilot project to investigate the use of whole-genome sequencing as a tool for clinical research. By piloting the acquisition of large amounts of DNA sequence data from individual human subjects, we are fostering the development of hypothesis-generating approaches for performing research in genomic medicine, including the exploration of issues related to the genetic architecture of disease, implementation of genomic technology, informed consent, disclosure of genetic information, and archiving, analyzing, and displaying sequence data. In the initial phase of ClinSeq, we are enrolling roughly 1000 participants; the evaluation of each includes obtaining a detailed family and medical history, as well as a clinical evaluation. The participants are being consented broadly for research on many traits and for whole-genome sequencing. Initially, Sanger-based sequencing of 300-400 genes thought to be relevant to atherosclerosis is being performed, with the resulting data analyzed for rare, high-penetrance variants associated with specific clinical traits. The participants are also being consented to allow the contact of family members for additional studies of sequence variants to explore their potential association with specific phenotypes. Here, we present the general considerations in designing ClinSeq, preliminary results based on the generation of an initial 826 Mb of sequence data, the findings for several genes that serve as positive controls for the project, and our views about the potential implications of ClinSeq. The early experiences with ClinSeq illustrate how large-scale medical sequencing can be a practical, productive, and critical component of research in genomic medicine.

Gray platelet syndrome: natural history of a large patient cohort and locus assignment to chromosome 3p.

Gray platelet syndrome (GPS) is an inherited bleeding disorder characterized by macrothrombocytopenia and absence of platelet α-granules resulting in typical gray platelets on peripheral smears. GPS is associated with a bleeding tendency, myelofibrosis, and splenomegaly. Reports on GPS are limited to case presentations. The causative gene and underlying pathophysiology are largely unknown. We present the results of molecular genetic analysis of 116 individuals including 25 GPS patients from 14 independent families as well as novel clinical data on the natural history of the disease. The mode of inheritance was autosomal recessive (AR) in 11 and indeterminate in 3 families. Using genome-wide linkage analysis, we mapped the AR-GPS gene to a 9.4-Mb interval on 3p21.1-3p22.1, containing 197 protein-coding genes. Sequencing of 1423 (69%) of the 2075 exons in the interval did not identify the GPS gene. Long-term follow-up data demonstrated the progressive nature of the thrombocytopenia and myelofibrosis of GPS resulting in fatal hemorrhages in some patients. We identified high serum vitamin B(12) as a consistent, novel finding in GPS. Chromosome 3p21.1-3p22.1 has not been previously linked to a platelet disorder; identification of the GPS gene will likely lead to the discovery of novel components of platelet organelle biogenesis. This study is registered at www.clinicaltrials.gov as NCT00069680 and NCT00369421.

Establishing analytical validity of BeadChip array genotype data by comparison to whole-genome sequence and standard benchmark datasets.

BACKGROUND: Clinical use of genotype data requires high positive predictive value (PPV) and thorough understanding of the genotyping platform characteristics. BeadChip arrays, such as the Global Screening Array (GSA), potentially offer a high-throughput, low-cost clinical screen for known variants. We hypothesize that quality assessment and comparison to whole-genome sequence and benchmark data establish the analytical validity of GSA genotyping. METHODS: To test this hypothesis, we selected 263 samples from Coriell, generated GSA genotypes in triplicate, generated whole genome sequence (rWGS) genotypes, assessed the quality of each set of genotypes, and compared each set of genotypes to each other and to the 1000 Genomes Phase 3 (1KG) genotypes, a performance benchmark. For 59 genes (MAP59), we also performed theoretical and empirical evaluation of variants deemed medically actionable predispositions. RESULTS: Quality analyses detected sample contamination and increased assay failure along the chip margins. Comparison to benchmark data demonstrated that > 82% of the GSA assays had a PPV of 1. GSA assays targeting transitions, genomic regions of high complexity, and common variants performed better than those targeting transversions, regions of low complexity, and rare variants. Comparison of GSA data to rWGS and 1KG data showed > 99% performance across all measured parameters. Consistent with predictions from prior studies, the GSA detection of variation within the MAP59 genes was 3/261. CONCLUSION: We establish the analytical validity of GSA assays using quality analytics and comparison to benchmark and rWGS data. GSA assays meet the standards of a clinical screen although assays interrogating rare variants, transversions, and variants within low-complexity regions require careful evaluation.

Personalized genomic medicine: lessons from the exome.

While genomic sequencing methods are powerful tools in the discovery of the genetic underpinnings of human disease, incidentally-revealed novel genomic risk factors may be equally important, both scientifically, and as relates to direct patient care. We performed whole-exome sequencing on a child with VACTERL association who suffered severe post-surgical neonatal pulmonary hypertension, and identified a potential novel genetic risk factor for this complication: a heterozygous mutation in CPSI. Newborn screening results from this patient's monozygotic twin provided evidence that this mutation, in combination with an environmental trigger (in this case, surgery), may have resulted in pulmonary artery hypertension due to inadequate nitric oxide production. Identification of this genetic risk factor allows for targeted medical preventative measures in this patient as well as relatives with the same mutation, and illustrates the power of incidental medical information unearthed by whole-exome sequencing.

Predicting protein domain interactions from coevolution of conserved regions.

The knowledge of protein and domain interactions provide crucial insights into their function within a cell. Several computational methods have been proposed to detect interactions between proteins and their constitutive domains. In this work, we focus on approaches based on correlated evolution (coevolution) of sequences of interacting proteins. In this type of approach, often referred to as the mirrortree method, a high correlation of evolutionary histories of two proteins is used as an indicator to predict protein interactions. Recently, it has been observed that subtracting the underlying speciation process by separating coevolution due to common speciation divergence from that due to common function of interacting pairs greatly improves the predictive power of the mirrortree approach. In this article, we investigate possible improvements and limitations of this method. In particular, we demonstrate that the performance of the mirrortree method that can be further improved by restricting the coevolution analysis to the relatively conserved regions in the protein domain sequences (disregarding highly divergent regions). We provide a theoretical validation of our results leading to new insights into the interplay between coevolution and speciation of interacting proteins.

Co-evolutionary analysis of domains in interacting proteins reveals insights into domain-domain interactions mediating protein-protein interactions.

Recent advances in functional genomics have helped generate large-scale high-throughput protein interaction data. Such networks, though extremely valuable towards molecular level understanding of cells, do not provide any direct information about the regions (domains) in the proteins that mediate the interaction. Here, we performed co-evolutionary analysis of domains in interacting proteins in order to understand the degree of co-evolution of interacting and non-interacting domains. Using a combination of sequence and structural analysis, we analyzed protein-protein interactions in F1-ATPase, Sec23p/Sec24p, DNA-directed RNA polymerase and nuclear pore complexes, and found that interacting domain pair(s) for a given interaction exhibits higher level of co-evolution than the non-interacting domain pairs. Motivated by this finding, we developed a computational method to test the generality of the observed trend, and to predict large-scale domain-domain interactions. Given a protein-protein interaction, the proposed method predicts the domain pair(s) that is most likely to mediate the protein interaction. We applied this method on the yeast interactome to predict domain-domain interactions, and used known domain-domain interactions found in PDB crystal structures to validate our predictions. Our results show that the prediction accuracy of the proposed method is statistically significant. Comparison of our prediction results with those from two other methods reveals that only a fraction of predictions are shared by all the three methods, indicating that the proposed method can detect known interactions missed by other methods. We believe that the proposed method can be used with other methods to help identify previously unrecognized domain-domain interactions on a genome scale, and could potentially help reduce the search space for identifying interaction sites.

CDD: a Conserved Domain Database for protein classification.

The Conserved Domain Database (CDD) is the protein classification component of NCBI's Entrez query and retrieval system. CDD is linked to other Entrez databases such as Proteins, Taxonomy and PubMed, and can be accessed at http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=cdd. CD-Search, which is available at http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi, is a fast, interactive tool to identify conserved domains in new protein sequences. CD-Search results for protein sequences in Entrez are pre-computed to provide links between proteins and domain models, and computational annotation visible upon request. Protein-protein queries submitted to NCBI's BLAST search service at http://www.ncbi.nlm.nih.gov/BLAST are scanned for the presence of conserved domains by default. While CDD started out as essentially a mirror of publicly available domain alignment collections, such as SMART, Pfam and COG, we have continued an effort to update, and in some cases replace these models with domain hierarchies curated at the NCBI. Here, we report on the progress of the curation effort and associated improvements in the functionality of the CDD information retrieval system.