The CDKN2A database: Integrating allelic variants with evolution, structure, function, and disease association

CDKN2A Germline Rare Coding Variants and Risk of Pancreatic Cancer in Minority Populations

Background: Pathogenic germline mutations in the CDKN2A tumor suppressor gene are rare and associated with highly penetrant familial melanoma and pancreatic cancer in non-Hispanic whites (NHW). To date, the prevalence and impact of CDKN2A rare coding variants (RCV) in racial minority groups remain poorly characterized. We examined the role of CDKN2A RCVs on the risk of pancreatic cancer among minority subjects.Methods: We sequenced CDKN2A in 220 African American (AA) pancreatic cancer cases, 900 noncancer AA controls, and 183 Nigerian controls. RCV frequencies were determined for each group and compared with that of 1,537 NHW patients with pancreatic cancer. Odds ratios (OR) and 95% confidence intervals (CI) were calculated for both a case-case comparison of RCV frequencies in AAs versus NHWs, and case-control comparison between AA cases versus noncancer AA controls plus Nigerian controls. Smaller sets of Hispanic and Native American cases and controls also were sequenced.Results: One novel missense RCV and one novel frameshift RCV were found among AA patients: 400G>A and 258_278del. RCV carrier status was associated with increased risk of pancreatic cancer among AA cases (11/220; OR, 3.3; 95% CI, 1.5-7.1; P = 0.004) compared with AA and Nigerian controls (17/1,083). Further, AA cases had higher frequency of RCVs: 5.0% (OR, 13.4; 95% CI, 4.9-36.7; P < 0.001) compared with NHW cases (0.4%).Conclusions: CDKN2A RCVs are more common in AA than in NHW patients with pancreatic cancer and associated with moderately increased pancreatic cancer risk among AAs.Impact: RCVs in CDKN2A are frequent in AAs and are associated with risk for pancreatic cancer. Cancer Epidemiol Biomarkers Prev; 27(11); 1364-70. ©2018 AACR.

The Human Variome Project: ensuring the quality of DNA variant databases in inherited renal disease

A recent review identified 60 common inherited renal diseases caused by DNA variants in 132 different genes. These diseases can be diagnosed with DNA sequencing, but each gene probably also has a thousand normal variants. Many more normal variants have been characterised by individual laboratories than are reported in the literature or found in publicly accessible collections. At present, testing laboratories must assess each novel change they identify for pathogenicity, even when this has been done elsewhere previously, and the distinction between normal and disease-associated variants is particularly an issue with the recent surge in exomic sequencing and gene discovery projects. The Human Variome Project recommends the establishment of gene-specific DNA variant databases to facilitate the sharing of DNA variants and decisions about likely disease causation. Databases improve diagnostic accuracy and testing efficiency, and reduce costs. They also help with genotype-phenotype correlations and predictive algorithms. The Human Variome Project advocates databases that use standardised descriptions, are up-to-date, include clinical information and are freely available. Currently, the genes affected in the most common inherited renal diseases correspond to 350 different variant databases, many of which are incomplete or have insufficient clinical details for genotype-phenotype correlations. Assistance is needed from nephrologists to maximise the usefulness of these databases for the diagnosis and management of inherited renal disease.

Mutation databases for inherited renal disease: are they complete, accurate, clinically relevant, and freely available?

This study examined whether gene-specific DNA variant databases for inherited diseases of the kidney fulfilled the Human Variome Project recommendations of being complete, accurate, clinically relevant and freely available. A recent review identified 60 inherited renal diseases caused by mutations in 132 genes. The disease name, MIM number, gene name, together with "mutation" or "database," were used to identify web-based databases. Fifty-nine diseases (98%) due to mutations in 128 genes had a variant database. Altogether there were 349 databases (a median of 3 per gene, range 0-6), but no gene had two databases with the same number of variants, and 165 (50%) databases included fewer than 10 variants. About half the databases (180, 54%) had been updated in the previous year. Few (77, 23%) were curated by "experts" but these included nine of the 11 with the most variants. Even fewer databases (41, 12%) included clinical features apart from the name of the associated disease. Most (223, 67%) could be accessed without charge, including those for 50 genes (40%) with the maximum number of variants. Future efforts should focus on encouraging experts to collaborate on a single database for each gene affected in inherited renal disease, including both unpublished variants, and clinical phenotypes.

Diagnostic value of DNA alteration: loss of heterozygosity or allelic imbalance-promising for molecular staging of prostate cancers

The biological behavior of prostate cancer is uncertain, and therefore, search for molecular prognostic markers associated with disease progression seems to be essential. We performed microsatellite allelotyping of DNA isolated from primary prostate tumors biopsies (prostate adenocarcinoma, PCa). We evaluated the frequency of allelic imbalance (AI), including loss of heterozygosity and/or microsatellite imbalance (LOH/MSI) as well as the association of these DNA alterations with clinicopathological variables. We assessed the significance of LOH/MSI alterations in selected imprinted and non-imprinted chromosomal regions (IR and NIR) in PCa. A total of 50 biopsies of prostate tumor (containing >75 % tumor cells) were histologically examined confirming prostate carcinoma. Microsatellite allelotyping using 16 microsatellite markers linked to the following chromosomal regions: 1p31.2, 3p21.3–25.3, 7q32.2, 9p21.3, 11p15.5, 12q23.2, and 16q22.1 was performed. The incidence of LOH/MSI alterations in prostate tumor cells was the highest for chromosomal regions 7q32.2 and 16q22.1 (31.25 and 26.60 %, respectively), followed by 1p31.2 and 3p21.3–25.3 (26.50 and 17.40 %, respectively). Statistically significant increase in LOH/MSI alterations has been observed for markers: D1S2137 (1p region; p = 0.00032), D9S974 (9p region; p = 0.0017), and D16S3025 (16q region; p = 0.0017). Statistically significant differences in frequency of LOH/MSI alterations in particular chromosomal regions have been found for 1p31.2, 7q32.2 and 16q22.1 (p = 0.027, p = 0.012 and p = 0.031, respectively). We documented statistically significant association between Fractional Allele Loss (FAL) index and advanced tumor stage (p < 0.05). We suggest that genomic instability of LOH/MSI type is a frequent event in prostate carcinogenesis and assessed as FAL index has clinical value for the molecular staging of prostate cancer in (TRUS)-guided prostate biopsy material.

p16 gene mutations in Barrett's esophagus in gastric metaplasia - intestinal metaplasia - dysplasia - adenocarcinoma sequence

PURPOSE

Barrett's associated esophageal adenocarcinoma (ADC) is one of the malignancies of most rapidly increasing incidence. The aim of the study was to assess p16 tumor suppressor gene alterations in the ADC premalignant conditions.

MATERIAL & METHODS

In the present study two p16 gene mutations (A148T and I49S) analysis with PCR- RFLP method have been performed in oesophageal biopsy specimen in 33 patients with Barrett's gastric metaplasia (GM), 27 - with Barrett's intestinal metaplasia (IM), 8 - with dysplasia and 11 - with ADC.

RESULTS

We have detected the I49S mutation in 12% (4/33) patients with GM, 18% (5/27) with IM, 50% - with dysplasia (4/8) and in 27% (3/11) - with ADC. The A148T mutation were found in 3% (1/33) patients with GM, 22% (6/27) - IM, 25% (2/8) - dysplasia and 27% patients with ADC (3/11). The frequency of the A148S mutation was rising in GM - IM - dysplasia - ADC sequence and was significantly lower in GM compared to all other grades taken together (p=0.0256). The frequency of the I49S mutation was rising in GM - IM - dysplasia sequence, to drop in ADC cases. There were no significant differences in frequency of the I49S mutation between studied groups.

CONCLUSIONS

These findings are consistent with the hypothesis on the role of the p16 mutations in early phase of Barrett's epithelium progression to ADC. The presence of p16 mutations in esophageal metaplastic columnar epithelium without goblet cells suggest that this pathology may have malignancy potential.

Classifying variants of CDKN2A using computational and laboratory studies

Variants in the CDKN2A tumor suppressor are associated with Familial Melanoma (FM), although for many variants the linkage is weak. The effects of missense variants on protein function and pathogenicity are often unclear. Multiple methods (e.g., laboratory, computational, epidemiological) have been developed to analyze whether a missense variant is pathogenic or not. It is not yet clear how to integrate these data types into a strategy for variant classification. We studied 51 CDKN2A missense variants using a cell cycle arrest assay. There was a continuum of results ranging from full wild-type effect through partial activity to complete loss of arrest. A reproducible decrease of 30% of cell cycle arrest activity correlated with FM association. We analyzed missense CDKN2A germline variants using a Bayesian method to combine multiple data types and derive a probability of pathogenicity. When equal to or more than two data types could be evaluated with this method, 22 of 25 FM-associated variants and 8 of 15 variants of uncertain significance were classified as likely pathogenic with >95% probability. The other 10 variants were classified as uncertain (probability 5-95%). For most variants, there were insufficient data to draw a conclusion. The Bayesian model appears to be a sound method of classifying missense variants in cancer susceptibility genes.

Curating gene variant databases (LSDBs): toward a universal standard

Gene variant databases or Locus-Specific DataBases (LSDBs) are used to collect and display information on sequence variants on a gene-by-gene basis. Their most frequent use is in relation to DNA-based diagnostics, giving clinicians and scientists easy access to an up-to-date overview of all gene variants identified worldwide and whether they influence the function of the gene ("pathogenic or not"). While literature on gene variant databases is extensive, little has been published on the process of database curation itself. Based on our extensive experience as LSDB curators and our contributions to database curation courses, we discuss the subject of database curation. We describe the tasks involved, the steps to take, and the issues that might occur. Our overview is a first step toward establishing overall guidelines for database curation and ultimately covers one aspect of establishing quality-assured gene variant databases.

ENIGMA--evidence-based network for the interpretation of germline mutant alleles: an international initiative to evaluate risk and clinical significance associated with sequence variation in BRCA1 and BRCA2 genes

As genetic testing for predisposition to human diseases has become an increasingly common practice in medicine, the need for clear interpretation of the test results is apparent. However, for many disease genes, including the breast cancer susceptibility genes BRCA1 and BRCA2, a significant fraction of tests results in the detection of a genetic variant for which disease association is not known. The finding of an "unclassified" variant (UV)/variant of uncertain significance (VUS) complicates genetic test reporting and counseling. As these variants are individually rare, a large collaboration of researchers and clinicians will facilitate studies to assess their association with cancer predisposition. It was with this in mind that the ENIGMA consortium (www.enigmaconsortium.org) was initiated in 2009. The membership is both international and interdisciplinary, and currently includes more than 100 research scientists and clinicians from 19 countries. Within ENIGMA, there are presently six working groups focused on the following topics: analysis, clinical, database, functional, tumor histopathology, and mRNA splicing. ENIGMA provides a mechanism to pool resources, exchange methods and data, and coordinately develop and apply algorithms for classification of variants in BRCA1 and BRCA2. It is envisaged that the research and clinical application of models developed by ENIGMA will be relevant to the interpretation of sequence variants in other disease genes.

Database resources for proteomics-based analysis of cancer

Biological/bioinformatics databases are essential for medical and biological studies. They integrate and organize biologically related information in a structured format and provide researchers with easy access to a variety of relevant data. This review presents an overview of publicly available databases relevant to proteomics studies in cancer research. They include gene/protein expression databases, gene mutation and single nucleotide polymorphisms databases, tumor antigen databases, protein-protein interaction, and biological pathway databases. Automated information retrieval from these databases enables efficient large-scale proteomics data analysis.

Potentially functional polymorphisms in cell cycle genes and the survival of non-small cell lung cancer in a Chinese population

The cell cycle governs the proliferation and growth of cells and is strictly controlled by some regulators including cyclins, CDKs and CKIs. Germ-line and somatic mutations in cell cycle genes were frequently observed in a subset of cancers including non-small cell lung cancer (NSCLC). In this study, we hypothesized that potentially functional single nucleotide polymorphisms (SNPs) in cell cycle genes may contribute to the prognosis of NSCLC in China. 54 potentially functional polymorphisms in key cell cycle genes (CDK1, CDK2, CDK4, CDK6, CDK7, CCND1, CCND2, CCND3, CCNE1, CCNA1, CCNA2, CCNB1, CCNH, p15, p16, p18, p19, p21, p27, Cdc25A and Cdc25B) were genotyped by using Illumina SNP genotyping platform to evaluate their associations with survival of NSCLC in a clinical cohort of 568 patients. We found that p18 rs3176447 variant genotypes were significantly associated with the decreased risk of death of NSCLC patients (adjusted HR=0.74, 95% CI=0.57-0.97 in an additive model; adjusted HR=0.76, 95% CI=0.55-0.97 in a dominant model); however, p21 rs2395655 variant genotypes were significantly associated with the increased risk of death (adjusted HR=1.21, 95% CI=1.02-1.42 in an additive model; adjusted HR=1.38, 95% CI=1.07-1.78 in a recessive model). Furthermore, the combined effect of unfavorable genotypes for these two SNPs was more prominent in patients with squamous cell carcinoma, late stage and without chemo- or radio-therapy. Although the exact biological function remains to be explored, our findings suggest possible association of polymorphisms of p18 and p21 with the prognosis of NSCLC in a Chinese population. Further large and functional studies are needed to confirm our findings.

p16 mutation spectrum in the premalignant condition Barrett's esophagus

Background

Mutation, promoter hypermethylation and loss of heterozygosity involving the tumor suppressor gene p16 (CDKN2a/INK4a) have been detected in a wide variety of human cancers, but much less is known concerning the frequency and spectrum of p16 mutations in premalignant conditions.

Methods and Findings

We have determined the p16 mutation spectrum for a cohort of 304 patients with Barrett's esophagus, a premalignant condition that predisposes to the development of esophageal adenocarcinoma. Forty seven mutations were detected by sequencing of p16 exon 2 in 44 BE patients (14.5%) with a mutation spectrum consistent with that caused by oxidative damage and chronic inflammation. The percentage of patients with p16 mutations increased with increasing histologic grade. In addition, samples from 3 out of 19 patients (15.8%) who underwent esophagectomy were found to have mutations.

Conclusions

The results of this study suggest the environment of the esophagus in BE patients can both generate and select for clones with p16 mutations.

Locus-specific databases and recommendations to strengthen their contribution to the classification of variants in cancer susceptibility genes

Locus-specific databases (LSDBs) are curated collections of sequence variants in genes associated with disease. LSDBs of cancer-related genes often serve as a critical resource to researchers, diagnostic laboratories, clinicians, and others in the cancer genetics community. LSDBs are poised to play an important role in disseminating clinical classification of variants. The IARC Working Group on Unclassified Genetic Variants has proposed a new system of five classes of variants in cancer susceptibility genes. However, standards are lacking for reporting and analyzing the multiple data types that assist in classifying variants. By adhering to standards of transparency and consistency in the curation and annotation of data, LSDBs can be critical for organizing our understanding of how genetic variation relates to disease. In this article we discuss how LSDBs can accomplish these goals, using existing databases for BRCA1, BRCA2, MSH2, MLH1, TP53, and CDKN2A to illustrate the progress and remaining challenges in this field. We recommend that: 1) LSDBs should only report a conclusion related to pathogenicity if a consensus has been reached by an expert panel. 2) The system used to classify variants should be standardized. The Working Group encourages use of the five class system described in this issue by Plon and colleagues. 3) Evidence that supports a conclusion should be reported in the database, including sources and criteria used for assignment. 4) Variants should only be classified as pathogenic if more than one type of evidence has been considered. 5) All instances of all variants should be recorded.

Genome wide identification of recessive cancer genes by combinatorial mutation analysis

We devised a novel procedure to identify human cancer genes acting in a recessive manner. Our strategy was to combine the contributions of the different types of genetic alterations to loss of function: amino-acid substitutions, frame-shifts, gene deletions. We studied over 20,000 genes in 3 Gigabases of coding sequences and 700 array comparative genomic hybridizations. Recessive genes were scored according to nucleotide mismatches under positive selective pressure, frame-shifts and genomic deletions in cancer. Four different tests were combined together yielding a cancer recessive p-value for each studied gene. One hundred and fifty four candidate recessive cancer genes (p-value<1.5×10⁻⁷, FDR = 0.39) were identified. Strikingly, the prototypical cancer recessive genes TP53, PTEN and CDKN2A all ranked in the top 0.5% genes. The functions significantly affected by cancer mutations are exactly overlapping those of known cancer genes, with the critical exception for the absence of tyrosine kinases, as expected for a recessive gene-set.

Genetic tests for common diseases: new insights, old concerns

The clinical utility of newly identified genetic variants associated with common diseases needs evaluation

Low prevalence of p53, p16(INK4a) and Ha-ras tumour-specific mutations in low-graded actinic keratosis

BACKGROUND

Ultraviolet radiation induces DNA damage and is the major risk factor for the development of non-melanoma skin cancer (NMSC). Different mutation rates of p53, p16(INK4a) and Ha-ras in cutaneous squamous cell carcinoma (SCC) and the earlier stage actinic keratosis (AK) have been reported.

OBJECTIVES

To assess the presence of missense mutations in hotspot exons of p53, p16(INK4a) and Ha-ras in low-graded AK.

PATIENTS/METHODS

Cryo-biopsies of 75 sun-exposed AK lesions and 75 sun-shielded areas of normal skin from 75 AK patients were analysed to identify mutations in p53 (exons 7 and 8), p16(INK4a) (exon 2) and Ha-ras (exon 1) using polymerase chain reaction (PCR) followed by direct sequencing. As a representative subset of the specimens, ten mutation-negative AK were also micro-dissected in order to exclude the possibility that additional mutations were undetected.

RESULTS

Eight missense and one nonsense point mutations were found in the 75 AK lesions examined (12%), of which seven (9%) were tumour-specific (i.e. present in AK lesions only) and two (3%) were p16(INK4a) mutations (i.e. also detected in normal skin). Three of the tumour-specific mutations (42%) were cytosine (C) to thymine (T) transitions at pyrimidine-rich sequences. Tumour-specific mutations were identified in 1% of p16(INK4a) (exon 2), 1% of Ha-ras (exon 1) and at a higher rate of 7% in p53 (exons 7 and 8), including one nonsense mutation.

CONCLUSIONS

The evaluation of a large number of AK specimens in this study have found a low gene mutation rate in low-graded AK lesions. p53 mutations rather than p16(INK4a) and/or Ha-ras mutations may be an early event in the development of AK to cutaneous SCC.

Recommendations for locus-specific databases and their curation

Expert curation and complete collection of mutations in genes that affect human health is essential for proper genetic healthcare and research. Expert curation is given by the curators of gene-specific mutation databases or locus-specific databases (LSDBs). While there are over 700 such databases, they vary in their content, completeness, time available for curation, and the expertise of the curator. Curation and LSDBs have been discussed, written about, and protocols have been provided for over 10 years, but there have been no formal recommendations for the ideal form of these entities. This work initiates a discussion on this topic to assist future efforts in human genetics. Further discussion is welcome.

Interpreting missense variants: comparing computational methods in human disease genes CDKN2A, MLH1, MSH2, MECP2, and tyrosinase (TYR)

The human genome contains frequent single-basepair variants that may or may not cause genetic disease. To characterize benign vs. pathogenic missense variants, numerous computational algorithms have been developed based on comparative sequence and/or protein structure analysis. We compared computational methods that use evolutionary conservation alone, amino acid (AA) change alone, and a combination of conservation and AA change in predicting the consequences of 254 missense variants in the CDKN2A (n = 92), MLH1 (n = 28), MSH2 (n = 14), MECP2 (n = 30), and tyrosinase (TYR) (n = 90) genes. Variants were validated as either neutral or deleterious by curated locus-specific mutation databases and published functional data. All methods that use evolutionary sequence analysis have comparable overall prediction accuracy (72.9-82.0%). Mutations at codons where the AA is absolutely conserved over a sufficient evolutionary distance (about one-third of variants) had a 91.6 to 96.8% likelihood of being deleterious. Three algorithms (SIFT, PolyPhen, and A-GVGD) that differentiate one variant from another at a given codon did not significantly improve predictive value over conservation score alone using the BLOSUM62 matrix. However, when all four methods were in agreement (62.7% of variants), predictive value improved to 88.1%. These results confirm a high predictive value for methods that use evolutionary sequence conservation, with or without considering protein structural change, to predict the clinical consequences of missense variants. The methods can be generalized across genes that cause different types of genetic disease. The results support the clinical use of computational methods as one tool to help interpret missense variants in genes associated with human genetic disease.

Ultraviolet radiation and melanoma: a systematic review and analysis of reported sequence variants

Despite the large body of mutational data available for melanoma and epidemiological studies linking this cancer to ultraviolet radiation (UVR), the fundamental carcinogenic mechanisms involved in melanoma remain largely unknown. To this end, we systematically reviewed, extracted, and analyzed mutational data from the extant melanoma literature in an effort to gain more insight into its early pathogenic events. We searched PubMed (1966-January 2006) using the words "mutation" AND "melanoma" in the title or abstract. Out of 2,095 returned results, there were 203 eligible studies that were subsequently analyzed. We cataloged 8,201 somatic and cultured melanoma specimens and annotated 2,041 reported somatic sequence variants. The single BRAF c.1799T>A (p.Val600Glu) alteration is the most prevalent variant while other A:T>T:A transversions were uncommon. Four highly-recurrent, non-ultraviolet B (UVB) changes account for most of the NRAS and BRAF variants. CDKN2A, PTEN/MMAC1, and TP53 harbored statistically higher rates of UVB signature changes (64.2%, 52.4%, and 69.2%, respectively) than oncogenic loci (NRAS: 15.3% and BRAF: 2.4%). More specifically, cutaneous melanomas showed a significantly higher proportion of UVB signature mutations at both TP53 and CDKN2A when compared to non-skin cancers using data from their respective locus-specific databases. Superficial spreading and nodular melanomas had the highest rates of BRAF (53.4%) and NRAS (28.0%) mutations. In melanoma, there is sufficient mutational evidence to support a role for direct UVB participation, especially at TP53 and CDKN2A. For oncogenes, the role for UVB is less clear since functionally-activating changes are uncommon and are subject to sequence constraints.

A common variant of the p16(INK4a) genetic region is associated with physical function in older people

p16(INK4a) is active in cell senescence, ageing and tumor suppression. Deletion of the small p16(INK4a)/ARF/p15(INK4b) region occurs in many cancers. We screened 25 common polymorphisms across the region and three related genes for associations with physical functioning in older people. In an initial sample of 938 (aged 65-80 years) from the EPIC study (Norfolk, UK), the rs2811712 SNP minor allele (located between the shared p16(INK4a)/ARF locus and p15(INK4b)) was associated with reduced physical impairment. This association remained after testing an additional 1319 EPIC-Norfolk samples (p-value=0.013, total n=2257), and on independent replication in the InCHIANTI study (n=709, p=0.015), and at one-sided significance in Iowa-EPESE (n=419, p=0.079). Overall (n=3372), the prevalence of severely limited physical function was 15.0% in common homozygotes and 7.0% in rare homozygotes (per minor allele odds ratio=1.48, 95% CI: 1.17-1.88, p=0.001, adjusted for age, sex and study). This estimate was similar excluding screening set 1 (OR=1.45, 95% CI: 1.09-1.92, p=0.010, n=2434). These findings require further replication, but provide the first direct evidence that the p16(INK4a)/ARF/p15(INK4b) genetic region and the senescence machinery are active in physical ageing in heterogeneous human populations. The mechanism involved may be via greater cellular restorative activity and reduced stem cell senescence.

An automated procedure to identify biomedical articles that contain cancer-associated gene variants

The proliferation of biomedical literature makes it increasingly difficult for researchers to find and manage relevant information. However, identifying research articles containing mutation data, a requisite first step in integrating large and complex mutation data sets, is currently tedious, time-consuming and imprecise. More effective mechanisms for identifying articles containing mutation information would be beneficial both for the curation of mutation databases and for individual researchers. We developed an automated method that uses information extraction, classifier, and relevance ranking techniques to determine the likelihood of MEDLINE abstracts containing information regarding genomic variation data suitable for inclusion in mutation databases. We targeted the CDKN2A (p16) gene and the procedure for document identification currently used by CDKN2A Database curators as a measure of feasibility. A set of abstracts was manually identified from a MEDLINE search as potentially containing specific CDKN2A mutation events. A subset of these abstracts was used as a training set for a maximum entropy classifier to identify text features distinguishing "relevant" from "not relevant" abstracts. Each document was represented as a set of indicative word, word pair, and entity tagger-derived genomic variation features. When applied to a test set of 200 candidate abstracts, the classifier predicted 88 articles as being relevant; of these, 29 of 32 manuscripts in which manual curation found CDKN2A sequence variants were positively predicted. Thus, the set of potentially useful articles that a manual curator would have to review was reduced by 56%, maintaining 91% recall (sensitivity) and more than doubling precision (positive predictive value). Subsequent expansion of the training set to 494 articles yielded similar precision and recall rates, and comparison of the original and expanded trials demonstrated that the average precision improved with the larger data set. Our results show that automated systems can effectively identify article subsets relevant to a given task and may prove to be powerful tools for the broader research community. This procedure can be readily adapted to any or all genes, organisms, or sets of documents.

Functional impacts of non-synonymous single nucleotide polymorphisms: selective constraint and structural environments

In this work, we studied the correlations between selective constraint, structural environments and functional impacts of non-synonymous single nucleotide polymorphisms (nsSNPs). We found that the relation between solvent accessibility and functional impacts of nsSNPs is not as simple as generally thought. Finer structural classifications need to be taken into account to reveal the complex relations between the characteristics of a structure environment and its influence on the functional impacts of nsSNPs. We introduced two parameters for each structural environment, consensus residue percentage and residue distribution distance, to characterize the selective constraint imposed by the environment. Both parameters significantly correlate with the functional bias of nsSNPs across the structural environments. This result shows that selective constraint underlies the bias of a structural environment towards a certain type of nsSNPs (disease-associated or benign).