Prioritization of candidate disease genes by topological similarity between disease and protein diffusion profiles
© Zhu et al.; licensee BioMed Central Ltd. 2013
Published: 10 April 2013
Identification of gene-phenotype relationships is a fundamental challenge in human health clinic. Based on the observation that genes causing the same or similar phenotypes tend to correlate with each other in the protein-protein interaction network, a lot of network-based approaches were proposed based on different underlying models. A recent comparative study showed that diffusion-based methods achieve the state-of-the-art predictive performance.
In this paper, a new diffusion-based method was proposed to prioritize candidate disease genes. Diffusion profile of a disease was defined as the stationary distribution of candidate genes given a random walk with restart where similarities between phenotypes are incorporated. Then, candidate disease genes are prioritized by comparing their diffusion profiles with that of the disease. Finally, the effectiveness of our method was demonstrated through the leave-one-out cross-validation against control genes from artificial linkage intervals and randomly chosen genes. Comparative study showed that our method achieves improved performance compared to some classical diffusion-based methods. To further illustrate our method, we used our algorithm to predict new causing genes of 16 multifactorial diseases including Prostate cancer and Alzheimer's disease, and the top predictions were in good consistent with literature reports.
Our study indicates that integration of multiple information sources, especially the phenotype similarity profile data, and introduction of global similarity measure between disease and gene diffusion profiles are helpful for prioritizing candidate disease genes.
Programs and data are available upon request.
Elucidating the relationship between human genetic diseases and their causal genes is an important emerging topic in current systematic biology. Understanding the inherited basis of these interactions could both improve medical care and better understand gene functions, interactions, and pathways. Typically, a disease is associated with a linkage interval of 0.5-10 cM on the chromosome if single nucleotide polymorphisms (SNPs) in this interval are correlated with an increased probability to have the disease [1–3]. Then, these linkage intervals define a set of (up to several hundreds) candidate disease-causing genes [4, 5].
With the rapid accumulation of different kinds of genomic data, a lot of computational methods for prioritizing candidate casual genes of a given phenotype were proposed at the beginning of the 21st century. These methods are largely based on the similarity of characteristics of disease genes, including sequence-based features [6–8], expression patterns [9–11], and functional annotation data [12, 13]. Despite their good performances, these methods suffer from some inherent limitations, e.g., the incomplete and false-positive disease-causal genes data, ambiguous boundary between different diseases, and highly heterogeneous of diseases.
Recently, network-based analysis showed that gene products related to the same disease are prone to physically interact with each other [14–17]. Based on this observation, a number of computational approaches have been proposed to predict associations between genes and diseases. These methods mainly begin with an artificial disease interval and test their ability to identity a real causing gene among a fixed number of nearby control genes. According to their underlying methodology, these methods can be loosely grouped into three categories . The first category is the linkage methods, which assumed that the direct interaction partners of a disease protein are likely to associate with the same disease phenotype. It was found that gene products interacted with a known disease protein were shown to be tenfold enriched in true disease-causing genes , so many researchers searched the PPI network for direct or indirect interacting partners of known disease genes to find new possible causing genes [10, 20, 21]. The second category is module-based methods, which are based the observation that gene products belonging to the same topological, functional or disease module have a high likelihood of being involved in the same disease. These methods inspect disease modules by graph partition algorithms and treat their members as potential disease genes . The last category consists of diffusion-based methods [23–26]. In these algorithms, 'random walkers' are released from the protein products of known disease genes, and then diffuse along the PPI network, with a certain probability to return the original nodes. Compared to linkage-based and module-based methods, diffusion-based methods used information encoded in the full network topology as well as the placement of all known disease genes. So a recent comparative research found that diffusion-based methods achieved the best predictive performance on the same data set .
In the present paper, we propose a new diffusion-based method to prioritize candidate disease genes. The diffusion profile of a disease was defined as the stationary distribution of all candidate genes in the PPI network under a random walk with restart where similarities between phenotypes are incorporated. Similarly, the diffusion profile of a gene was obtained by smoothing the probability distribution over the whole network when starting a walk from this gene. Then, candidate disease genes are prioritized by comparing their diffusion profiles with that of the disease, measured by the linear correlation coefficient and cosine of the angle between profile vectors. Finally, the effectiveness of our method was demonstrated through the leave-one-out cross-validation against control genes from artificial linkage intervals and randomly chosen genes. Comparisons of our method with two classical diffusion-based methods showed that our method achieves improved performance. To further illustrate our method, we also used our algorithm to predict new causing genes of 16 multifactorial diseases including prostate cancer and Alzheimer's disease, and our top predictions are in good consistent with literature reports.
Protein-protein interaction data and known disease-gene associations
The protein-protein interaction network (PPI) is modelled as an undirected graph with nodes representing the genes and edges representing the physical or binding interactions between proteins encoded by the genes. In the present paper, PPI network is obtained from release 9 of the Human Protein Reference Database (HPRD) . After removing duplications and self-linked interactions, we obtain 37 064 manually curated interactions between 9515 human genes.
Disease-gene association data are downloaded from the Online Mendelian Inheritance in Man (OMIM) knowledgebase . The dataset contains 2704 diseases and 5316 disease-gene associations after removing the duplications, with an average of 1.97 gene associations for each disease. To facilitate the cross-validation, diseases currently associated with only one causal gene were discarded. Meanwhile, associations not correlated with the 9515 human genes in the PPI network were also excluded. After these steps, a total number of 1238 validate disease-gene associations are left for further consideration.
Random walk with restart and diffusion profile of a protein
Given a PPI network where is the set of proteins, is the set of interactions. The random walk on PPI network is defined as an iterative walker's transition from its current node to any neighbouring node with equal probability starting at a given source node(s). In a statistical point of view, a random walk is a finite Markov chain that is time-reversible.
which is the random walk with restart. The restart probability enforces a restriction on how far we want the random walker to get away from the start node . If is close to 1, the stationary probability vector reflects the local structure around the , and as gets close to 0, a more global view is observed. Finally, we define the diffusion profile of a protein g as the stationary distribution of the random walk with the initial distribution.
Random walk with restart (RWR) provides a good relevance between two nodes in a graph, and it has been successfully used in numerous settings, like automatic captioning of images, generalizations to the "connection subgraphs", personalized PageRank, and so on . The main advantage of the random walk method is that its computational complexity is relatively low and applicable to handle large PPI networks. Moreover, the method can be used to compute the proximity of a node to a set of source nodes (not just a single source node). This property is especially beneficial when a core set of members of a phenotype is known and the network is queried for candidate members.
Diffusion profile of a disease by incorporating OMIM Phenotype similarity and prioritizing function
In this way, proteins that interact with several disease proteins will gain a high probabilistic weight, as well as those that may not directly interact with any disease proteins but are in close network proximity to them.
where , is the initial distribution of disease .
We called the stationary distribution of the random walk diffusion profile of the disease d.
If a gene has very similar stationary distribution profile with a disease, it may have strong evidence to be the causing gene of the disease. Following this observation, given a disease d, its candidate genes were ranked according to the LCC and COS values between their stationary distributions. We referred the two proposed candidate gene prioritization algorithms as DP_LCC and DP_COS respectively, where 'DP' is the abbreviation of 'diffusion profile'.
Cross-validation and evaluation criteria
We used two leave-one-out cross-validation methods to validate our algorithm. First is the artificial linkage interval approach, which assumes the singled out interaction is unknown and prioritizes the gene against a set of control genes in the genome. Here the control set consists of the nearest 99 genes around real disease causing genes according to the UCSC refGene table. Actually, there may be few undiscovered disease causing genes in the control set. Second, we used validation against random genes, i.e., in each run, a known disease-gene association is singled out as the test sample against a set of 99 control genes that are selected at random from all genes in the interactome. So, a total of 100 genes (including the real disease-causing gene) are served as test data, and performance of our method is validated by capability to recover the real causing gene from the rest 99 control genes.
We used two measures to evaluate the performance of the proposed method. For each cross-validation run, we calculated the proportion of disease genes that obtain the top prioritization score against the corresponding 99 control genes, and called this measure precision (PRE). Also, given a threshold of rank ratio, we calculated the sensitivity (also called the true positive rate) as the fraction of disease genes ranked above this threshold and the specificity (also called the true negative rate) as the fraction of control genes ranked below the threshold. Varying the threshold of rank ratio from 0 to 1 with the scale 0.01, we are able to draw a receiver operating characteristic (ROC) curve and further calculate the area under this curve (AUC). Clearly, a larger PRE/AUC values indicate a better prediction performance of a prioritization method.
Results and Discussion
Effects of parameters
Prediction accuracies of DP_LCC and DP_COS at = 0.25
The parameter controls the contribution of other related phenotypes to the initial distribution of a phenotype. Large introduce more global dependence of ranking between different correlated phenotypes. When , the proposed method could be very similar to the RWR algorithm. To investigate the effect of this parameter, we set various values of ranging from 0.1 to 0.9, the performance of our algorithm measured by two evaluation criteria are shown in Table 1. The performances of our algorithm evaluated by two criteria, i.e., LCC and COS, have no significant difference at different values of . The performance is improved when ranges from 0.1 to 0.5, and decreased when is larger than 0.6, especially at 1.0. Therefore, we suggest the value of 0.4 or 0.5.
Comparison with other diffusion-based methods
To illustrate the utility of the present method, we compared the performances of DP_LCC and DP_COS with two diffusion-based methods, i.e, the RWR and PRINCE on the same gene-phenotype network. Both methods used random walk with restart algorithm to prioritize disease-candidate genes, and achieved relatively better performance compared to linkage-based methods and graph partitioning-based methods . The only difference between RWR and PRINCE is the construction of initial distribution of a disease, where the initial probability vector of RWR was constructed such that equal probability was assigned to each causing gene of a disease, and in PRINCE, the prior information vector was initialized by incorporating disease similarity information by using a logistic function. In our implement, two free parameters c and d in logistic function are set to -15 and log(9999) respectively, which are in accordance with PRINCE.
Performances of different algorithms at different values of
The present work a
Predict novel causing genes of Prostate cancer and Alzheimer's disease
Top 10 predicted causal genes of 16 multifactoral diseases predicted by DP-LCC
Top-10 predictions for each phenotype by our algorithm
Alzheimer's disease (AD) is the most common cause of dementia in the elderly. It is characterized clinically by progressive memory loss that leads eventually to dementia. As is shown in Table 3, the third prediction for AD is TREM2 (Triggering Receptor Expressed on Myeloid cells), which is a member of the innate immune receptor TREM family [35, 36]. It is expressed on the cell surface of the monocyte-macrophage lineage including monocyte derived dendritic cells, osteoclasts and microglia in the Central Nervous System (CNS) [37, 38]. Recent researches showed that TREM2 deficiency originates a genetic syndrome characterized by bone cysts and presenile dementia . Another prediction, MAPT, was also a suspicious driver gene for AD [40–43]. Genetic variability at the MAPT locus was shown to be associated with increased risk for the sporadic tauopathies, PSP  and corticobasal degeneration . The fifth prediction is PSEN1, which is also a driver gene of AD in the literature. It was reported that mutations in the human presenilin genes (PSEN1 and PSEN2) are associated with early onset familial Alzheimer disease .
Diabetes is a chronic condition associated with abnormally high levels of sugar (glucose) in the blood. The disease can be classified into three different categories: the type I, type II and the gestational diabetes. The top 3 predictions of our algorithm for Diabetes mellitus are ABCG8, ABCG5 and PPP1R3A, respectively. The first two genes, i.e., ABCG5 and ABCE8, are ATP-binding cassette transporters that are located in a head-to-head orientation on chromosome 2. The proteins are expressed in the liver, intestine [47, 48], and gallbladder epithelial cells . Polymorphisms in ABCG5/ABCG8 genes might contribute to the genetic variation in plasma lipid levels and in cholesterol saturation of the bile . Down-regulation of hepatic and intestinal Abcg5 and Abcg8 expression associated with altered sterol fluxes in rats with streptozotocin-induced diabetes . In addition, defects in Rp1 and PPP1R3A are also causes of susceptibility to diabetes mellitus of type I and II, respectively .
Prostate cancer is the most common malignancy in men and the second leading cause of male cancer-related deaths in the Western world. According to the OMIM record, prostate cancer has 25 validate causing genes. Based on these 25 known genes and causing genes of textual related phenotypes, we predicted novel causing genes of prostate cancer using our method (DP_LCC). As is shown in Table 3, the top 3 predictions for prostate cancer are TP53, RET and DHCR24, where TP53 is an important suppressor involved in several types of cancer [52, 53]. According to the IARC TP53 Mutation Database , inactivating TP53 mutations are detected at frequencies in the range of 10-20% in primary prostate cancer . TP53 was also predicted as the tops by PRINCE. Our second prediction for prostate cancer is RET, which was also found to be overexpressed in high-grade (histopathologically advanced) prostatic intraepithelial neoplasia (PIN) and prostate cancer . So RET was supposed to play a role in the growth of both benign and neoplastic prostate epithelial cells. Another two predicted causing genes, DHCR24 and STK11, were also consistent with the literature. Specifically, DHCR24 is one of androgen receptor-regulated genes implicated in prostate carcinogenesis , and STK11 was reported to be inactivated in prostate cancer, through mutation analysis of 24 known cancer genes in the NCI-60 cell line set .
In this paper, a diffusion-based method incorporating pairwise similarities of phenotypes was proposed to prioritize candidate disease genes. The novelty of our method lies in the incorporation of disease phenotypes (OMIM phenotype data) from the literature to the initial state of the RWR, and the usage of global similarity between diffusion profiles of disease and genes. Diffusion profiles of diseases and genes are obtained by walking over the protein-protein interaction network under a given initial distribution, where the initial distribution of a disease was weighted by OMIM Phenotype similarities exceeding a threshold. Then the linear correlation coefficient and cosine of the angle between profiles of a disease and given genes were computed to rank the priorities with the disease. Leave-one-out cross-validation on a benchmark dataset showed that our method achieved a higher precision (measured by PRE and AUC) than existing diffusion-based methods. This result suggests that the proposed algorithm effectively captures the interplay between gene network and phenotype network. We finally predicted causing genes of 16 multifactorial diseases including Prostate cancer and Alzheimer's disease using our algorithm and found that parts of our predictions are in good accordance with current experimental reports.
The superior performance of our method was attributed to the following aspects. First, integration of multiple information sources, especially the phenotype similarity profile data. Second, global similarity measures (linear correlation coefficient and cosine function of the angle) between diffusion profiles of diseases and genes are introduced to prioritize candidate disease genes. In contrast to previous methods that prioritize candidate disease genes through comparing their corresponding components in the diffusion profile of a disease, our global-based method could take into consideration the distribution values of other genes in the PPI network,
Consequently, in the future work, we can integrate some other genomic information to further improve our method, such as gene expression data, functional annotations, pathway membership and so on. Moreover, many researchers pointed out that some diseases might be attributed to a certain protein complexes composed by multiple proteins or a certain pathways. So furthermore attention should be paid on elucidating associations between diseases and protein complexes or pathways.
This work was partially supported by the National Natural Science Foundation of China (No. 11171224 and 31100953), Shanghai Leading Academic Discipline Project (No. S30405) and Innovation Program of Shanghai Municipal Education Commission (No. 12YZ088) and the Program of Shanghai Normal University (DZL121).
This article has been published as part of BMC Bioinformatics Volume 14 Supplement 5, 2013: Proceedings of the Third Annual RECOMB Satellite Workshop on Massively Parallel Sequencing (RECOMB-seq 2013). The full contents of the supplement are available online at http://www.biomedcentral.com/bmcbioinformatics/supplements/14/S5.
The National Natural Science Foundation of China (No. 11171224 and 31100953) provided the funding for publication of this article.
- Birnbaum S, Ludwig KU, Reutter H, Herms S, Steffens M, Rubini M, Baluardo C, Ferrian M, Almeida de Assis N, Alblas MA: Key susceptibility locus for nonsyndromic cleft lip with or without cleft palate on chromosome 8q24. Nat Genet. 2009, 41 (4): 473-477. 10.1038/ng.333.View ArticlePubMedGoogle Scholar
- Kathiresan S, Voight BF, Purcell S, Musunuru K, Ardissino D, Mannucci PM, Anand S, Engert JC, Samani NJ, Schunkert H: Genome-wide association of early-onset myocardial infarction with single nucleotide polymorphisms and copy number variants. Nat Genet. 2009, 41 (3): 334-341. 10.1038/ng.327.View ArticlePubMedGoogle Scholar
- Liao B, Li X, Zhu W, Li R, Wang S: Multiple ant colony algorithm method for selecting tag SNPs. J Biomed Inform. 2012Google Scholar
- Anne W, van Rensburg H, Adams J, Ector H, Van de Werf F, Heidbuchel H: Ablation of post-surgical intra-atrial reentrant tachycardia. Predilection target sites and mapping approach. Eur Heart J. 2002, 23 (20): 1609-1616. 10.1053/euhj.2002.3168.View ArticlePubMedGoogle Scholar
- Botstein D, Risch N: Discovering genotypes underlying human phenotypes: past successes for mendelian disease, future approaches for complex disease. Nat Genet. 2003, 228-237. 33 SupplGoogle Scholar
- Adie EA, Adams RR, Evans KL, Porteous DJ, Pickard BS: SUSPECTS: enabling fast and effective prioritization of positional candidates. Bioinformatics. 2006, 22 (6): 773-774. 10.1093/bioinformatics/btk031.View ArticlePubMedGoogle Scholar
- Lopez-Bigas N, Ouzounis CA: Genome-wide identification of genes likely to be involved in human genetic disease. Nucleic Acids Res. 2004, 32 (10): 3108-3114. 10.1093/nar/gkh605.PubMed CentralView ArticlePubMedGoogle Scholar
- Turner FS, Clutterbuck DR, Semple CAM: POCUS: mining genomic sequence annotation to predict disease genes. Genome Biol. 2003, 4 (11):Google Scholar
- Aerts S, Lambrechts D, Maity S, Van Loo P, Coessens B, De Smet F, Tranchevent LC, De Moor B, Marynen P, Hassan B: Gene prioritization through genomic data fusion. Nature Biotechnology. 2006, 24 (5): 537-544. 10.1038/nbt1203.View ArticlePubMedGoogle Scholar
- Franke L, van Bakel H, Fokkens L, de Jong ED, Egmont-Petersen M, Wijmenga C: Reconstruction of a functional human gene network, with an application for prioritizing positional candidate genes. Am J Hum Genet. 2006, 78 (6): 1011-1025. 10.1086/504300.PubMed CentralView ArticlePubMedGoogle Scholar
- van Driel MA, Cuelenaere K, Kemmeren PP, Leunissen JA, Brunner HG, Vriend G: GeneSeeker: extraction and integration of human disease-related information from web-based genetic databases. Nucleic Acids Res. 2005, 33 (Web Server): W758-761. 10.1093/nar/gki435.PubMed CentralView ArticlePubMedGoogle Scholar
- Freudenberg J, Propping P: A similarity-based method for genome-wide prediction of disease-relevant human genes. Bioinformatics. 2002, 18 (Suppl 2): S110-115. 10.1093/bioinformatics/18.suppl_2.S110.View ArticlePubMedGoogle Scholar
- Perez-Iratxeta C, Bork P, Andrade MA: Association of genes to genetically inherited diseases using data mining. Nat Genet. 2002, 31 (3): 316-319.PubMedGoogle Scholar
- Goh KI, Cusick ME, Valle D, Childs B, Vidal M, Barabasi AL: The human disease network. P Natl Acad Sci USA. 2007, 104 (21): 8685-8690. 10.1073/pnas.0701361104.View ArticleGoogle Scholar
- Ideker T, Sharan R: Protein networks in disease. Genome Res. 2008, 18 (4): 644-652. 10.1101/gr.071852.107.PubMed CentralView ArticlePubMedGoogle Scholar
- Oti M, Brunner HG: The modular nature of genetic diseases. Clin Genet. 2007, 71 (1): 1-11.View ArticlePubMedGoogle Scholar
- Wu X, Jiang R, Zhang MQ, Li S: Network-based global inference of human disease genes. Mol Syst Biol. 2008, 4: 189-PubMed CentralView ArticlePubMedGoogle Scholar
- Barabasi AL, Gulbahce N, Loscalzo J: Network medicine: a network-based approach to human disease. Nat Rev Genet. 2011, 12 (1): 56-68. 10.1038/nrg2918.PubMed CentralView ArticlePubMedGoogle Scholar
- Oti M, Snel B, Huynen MA, Brunner HG: Predicting disease genes using protein-protein interactions. J Med Genet. 2006, 43 (8): 691-698. 10.1136/jmg.2006.041376.PubMed CentralView ArticlePubMedGoogle Scholar
- Krauthammer M, Kaufmann CA, Gilliam TC, Rzhetsky A: Molecular triangulation: bridging linkage and molecular-network information for identifying candidate genes in Alzheimer's disease. Proc Natl Acad Sci USA. 2004, 101 (42): 15148-15153. 10.1073/pnas.0404315101.PubMed CentralView ArticlePubMedGoogle Scholar
- Iossifov I, Zheng T, Baron M, Gilliam TC, Rzhetsky A: Genetic-linkage mapping of complex hereditary disorders to a whole-genome molecular-interaction network. Genome Res. 2008, 18 (7): 1150-1162. 10.1101/gr.075622.107.PubMed CentralView ArticlePubMedGoogle Scholar
- Navlakha S, Schatz MC, Kingsford C: Revealing biological modules via graph summarization. J Comput Biol. 2009, 16 (2): 253-264. 10.1089/cmb.2008.11TT.View ArticlePubMedGoogle Scholar
- Kohler S, Bauer S, Horn D, Robinson PN: Walking the interactome for prioritization of candidate disease genes. Am J Hum Genet. 2008, 82 (4): 949-958. 10.1016/j.ajhg.2008.02.013.PubMed CentralView ArticlePubMedGoogle Scholar
- Vanunu O, Magger O, Ruppin E, Shlomi T, Sharan R: Associating genes and protein complexes with disease via network propagation. PLoS Comput Biol. 2010, 6 (1): e1000641-10.1371/journal.pcbi.1000641.PubMed CentralView ArticlePubMedGoogle Scholar
- Li Y, Patra JC: Genome-wide inferring gene-phenotype relationship by walking on the heterogeneous network. Bioinformatics. 2010, 26 (9): 1219-1224. 10.1093/bioinformatics/btq108.View ArticlePubMedGoogle Scholar
- Hwang T, Zhang W, Xie M, Liu J, Kuang R: Inferring disease and gene set associations with rank coherence in networks. Bioinformatics. 2011, 27 (19): 2692-2699. 10.1093/bioinformatics/btr463.View ArticlePubMedGoogle Scholar
- Navlakha S, Kingsford C: The power of protein interaction networks for associating genes with diseases. Bioinformatics. 2010, 26 (8): 1057-1063. 10.1093/bioinformatics/btq076.PubMed CentralView ArticlePubMedGoogle Scholar
- Peri S, Navarro JD, Kristiansen TZ, Amanchy R, Surendranath V, Muthusamy B, Gandhi TK, Chandrika KN, Deshpande N, Suresh S: Human protein reference database as a discovery resource for proteomics. Nucleic Acids Res. 2004, 32 (Database): D497-501.PubMed CentralView ArticlePubMedGoogle Scholar
- Hamosh A, Scott AF, Amberger JS, Bocchini CA, McKusick VA: Online Mendelian Inheritance in Man (OMIM), a knowledgebase of human genes and genetic disorders. Nucleic Acids Res. 2005, 33 (Database): D514-517.PubMed CentralPubMedGoogle Scholar
- van Driel MA, Bruggeman J, Vriend G, Brunner HG, Leunissen JA: A text-mining analysis of the human phenome. Eur J Hum Genet. 2006, 14 (5): 535-542. 10.1038/sj.ejhg.5201585.View ArticlePubMedGoogle Scholar
- Lowe HJ, Barnett GO: Understanding and using the medical subject headings (MeSH) vocabulary to perform literature searches. JAMA. 1994, 271 (14): 1103-1108. 10.1001/jama.1994.03510380059038.View ArticlePubMedGoogle Scholar
- Tong HH, Faloutsos C, Pan JY: Random walk with restart: fast solutions and applications. Knowl Inf Syst. 2008, 14 (3): 327-346. 10.1007/s10115-007-0094-2.View ArticleGoogle Scholar
- Chen Y, Jiang T, Jiang R: Uncover disease genes by maximizing information flow in the phenome-interactome network. Bioinformatics. 2011, 27 (13): i167-176. 10.1093/bioinformatics/btr213.PubMed CentralView ArticlePubMedGoogle Scholar
- Wu XB, Jiang R, Zhang MQ, Li S: Network-based global inference of human disease genes. Mol Syst Biol. 2008, 4:Google Scholar
- Bouchon A, Facchetti F, Weigand MA, Colonna M: TREM-1 amplifies inflammation and is a crucial mediator of septic shock. Nature. 2001, 410 (6832): 1103-1107. 10.1038/35074114.View ArticlePubMedGoogle Scholar
- Daws MR, Lanier LL, Seaman WE, Ryan JC: Cloning and characterization of a novel mouse myeloid DAP12-associated receptor family. Eur J Immunol. 2001, 31 (3): 783-791. 10.1002/1521-4141(200103)31:3<783::AID-IMMU783>3.0.CO;2-U.View ArticlePubMedGoogle Scholar
- Colonna M: TREMs in the immune system and beyond. Nat Rev Immunol. 2003, 3 (6): 445-453. 10.1038/nri1106.View ArticlePubMedGoogle Scholar
- Schmid CD, Sautkulis LN, Danielson PE, Cooper J, Hasel KW, Hilbush BS, Sutcliffe JG, Carson MJ: Heterogeneous expression of the triggering receptor expressed on myeloid cells-2 on adult murine microglia. J Neurochem. 2002, 83 (6): 1309-1320. 10.1046/j.1471-4159.2002.01243.x.PubMed CentralView ArticlePubMedGoogle Scholar
- Fenoglio C, Galimberti D, Piccio L, Scalabrini D, Panina P, Buonsanti C, Venturelli E, Lovati C, Forloni G, Mariani C: Absence of TREM2 polymorphisms in patients with Alzheimer's disease and Frontotemporal Lobar Degeneration. Neurosci Lett. 2007, 411 (2): 133-137. 10.1016/j.neulet.2006.10.029.View ArticlePubMedGoogle Scholar
- Lilius L, Froelich Fabre S, Basun H, Forsell C, Axelman K, Mattila K, Andreadis A, Viitanen M, Winblad B, Fratiglioni L: Tau gene polymorphisms and apolipoprotein E epsilon4 may interact to increase risk for Alzheimer's disease. Neurosci Lett. 1999, 277 (1): 29-32. 10.1016/S0304-3940(99)00833-2.View ArticlePubMedGoogle Scholar
- Bullido MJ, Aldudo J, Frank A, Coria F, Avila J, Valdivieso F: A polymorphism in the tau gene associated with risk for Alzheimer's disease. Neurosci Lett. 2000, 278 (12): 49-52.View ArticlePubMedGoogle Scholar
- Roks G, Dermaut B, Heutink P, Julliams A, Backhovens H, Van de Broeck M, Serneels S, Hofman A, Van Broeckhoven C, van Duijn CM: Mutation screening of the tau gene in patients with early-onset Alzheimer's disease. Neurosci Lett. 1999, 277 (2): 137-139. 10.1016/S0304-3940(99)00861-7.View ArticlePubMedGoogle Scholar
- Russ C, Powell JF, Zhao J, Baker M, Hutton M, Crawford F, Mullan M, Roks G, Cruts M, Lovestone S: The microtubule associated protein Tau gene and Alzheimer's disease--an association study and meta-analysis. Neurosci Lett. 2001, 314 (12): 92-96.View ArticlePubMedGoogle Scholar
- Baker M, Litvan I, Houlden H, Adamson J, Dickson D, Perez-Tur J, Hardy J, Lynch T, Bigio E, Hutton M: Association of an extended haplotype in the tau gene with progressive supranuclear palsy. Hum Mol Genet. 1999, 8 (4): 711-715. 10.1093/hmg/8.4.711.View ArticlePubMedGoogle Scholar
- Houlden H, Baker M, Morris HR, MacDonald N, Pickering-Brown S, Adamson J, Lees AJ, Rossor MN, Quinn NP, Kertesz A: Corticobasal degeneration and progressive supranuclear palsy share a common tau haplotype. Neurology. 2001, 56 (12): 1702-1706. 10.1212/WNL.56.12.1702.View ArticlePubMedGoogle Scholar
- Rogaev EI, Sherrington R, Wu C, Levesque G, Liang Y, Rogaeva EA, Ikeda M, Holman K, Lin C, Lukiw WJ: Analysis of the 5' sequence, genomic structure, and alternative splicing of the presenilin-1 gene (PSEN1) associated with early onset Alzheimer disease. Genomics. 1997, 40 (3): 415-424. 10.1006/geno.1996.4523.View ArticlePubMedGoogle Scholar
- Tauscher A, Kuver R: ABCG5 and ABCG8 are expressed in gallbladder epithelial cells. Biochem Biophys Res Commun. 2003, 307 (4): 1021-1028. 10.1016/S0006-291X(03)01296-8.View ArticlePubMedGoogle Scholar
- Klett EL, Lee MH, Adams DB, Chavin KD, Patel SB: Localization of ABCG5 and ABCG8 proteins in human liver, gall bladder and intestine. BMC Gastroenterol. 2004, 4: 21-10.1186/1471-230X-4-21.PubMed CentralView ArticlePubMedGoogle Scholar
- Acalovschi M, Ciocan A, Mostean O, Tirziu S, Chiorean E, Keppeler H, Schirin-Sokhan R, Lammert F: Are plasma lipid levels related to ABCG5/ABCG8 polymorphisms? A preliminary study in siblings with gallstones. Eur J Intern Med. 2006, 17 (7): 490-494. 10.1016/j.ejim.2006.04.012.View ArticlePubMedGoogle Scholar
- Bloks VW, Bakker-Van Waarde WM, Verkade HJ, Kema IP, Wolters H, Vink E, Groen AK, Kuipers F: Down-regulation of hepatic and intestinal Abcg5 and Abcg8 expression associated with altered sterol fluxes in rats with streptozotocin-induced diabetes. Diabetologia. 2004, 47 (1): 104-112. 10.1007/s00125-003-1261-y.View ArticlePubMedGoogle Scholar
- Maier LM, Chapman J, Howson JM, Clayton DG, Pask R, Strachan DP, McArdle WL, Twells RC, Todd JA: No evidence of association or interaction between the IL4RA, IL4, and IL13 genes in type 1 diabetes. Am J Hum Genet. 2005, 76 (3): 517-521. 10.1086/428387.PubMed CentralView ArticlePubMedGoogle Scholar
- Soussi T, Beroud C: Assessing TP53 status in human tumours to evaluate clinical outcome. Nat Rev Cancer. 2001, 1 (3): 233-240.View ArticlePubMedGoogle Scholar
- Caron de Fromentel C, Soussi T: TP53 tumor suppressor gene: a model for investigating human mutagenesis. Genes Chromosomes Cancer. 1992, 4 (1): 1-15. 10.1002/gcc.2870040102.View ArticlePubMedGoogle Scholar
- Hernandez-Boussard T, Rodriguez-Tome P, Montesano R, Hainaut P: IARC p53 mutation database: a relational database to compile and analyze p53 mutations in human tumors and cell lines. International Agency for Research on Cancer. Hum Mutat. 1999, 14 (1): 1-8. 10.1002/(SICI)1098-1004(1999)14:1<1::AID-HUMU1>3.0.CO;2-H.View ArticlePubMedGoogle Scholar
- Muhlbauer KR, Grone HJ, Ernst T, Grone E, Tschada R, Hergenhahn M, Hollstein M: Analysis of human prostate cancers and cell lines for mutations in the TP53 and KLF6 tumour suppressor genes. Br J Cancer. 2003, 89 (4): 687-690. 10.1038/sj.bjc.6601164.PubMed CentralView ArticlePubMedGoogle Scholar
- Dawson DM, Lawrence EG, MacLennan GT, Amini SB, Kung HJ, Robinson D, Resnick MI, Kursh ED, Pretlow TP, Pretlow TG: Altered expression of RET proto-oncogene product in prostatic intraepithelial neoplasia and prostate cancer. J Natl Cancer Inst. 1998, 90 (7): 519-523. 10.1093/jnci/90.7.519.View ArticlePubMedGoogle Scholar
- Dong Y, Zhang H, Gao AC, Marshall JR, Ip C: Androgen receptor signaling intensity is a key factor in determining the sensitivity of prostate cancer cells to selenium inhibition of growth and cancer-specific biomarkers. Mol Cancer Ther. 2005, 4 (7): 1047-1055. 10.1158/1535-7163.MCT-05-0124.View ArticlePubMedGoogle Scholar
- Ikediobi ON, Davies H, Bignell G, Edkins S, Stevens C, O'Meara S, Santarius T, Avis T, Barthorpe S, Brackenbury L: Mutation analysis of 24 known cancer genes in the NCI-60 cell line set. Mol Cancer Ther. 2006, 5 (11): 2606-2612. 10.1158/1535-7163.MCT-06-0433.PubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.