Sihua Peng

Molecular classification of cancer types from microarray data using the combination of genetic algorithms and support vector machines

Simultaneous multiclass classification of tumor types is essential for future clinical implementations of microarray‐based cancer diagnosis. In this study, we have combined genetic algorithms (GAs) and all paired support vector machines (SVMs) for multiclass cancer identification. The predictive features have been selected through iterative SVMs/GAs, and recursive feature elimination post‐processing steps, leading to a very compact cancer‐related predictive gene set. Leave‐one‐out cross‐validations yielded accuracies of 87.93% for the eight‐class and 85.19% for the fourteen‐class cancer classifications, outperforming the results derived from previously published methods.

Multiclass cancer classification and biomarker discovery using GA-based algorithms

MOTIVATION The development of microarray-based high-throughput gene profiling has led to the hope that this technology could provide an efficient and accurate means of diagnosing and classifying tumors, as well as predicting prognoses and effective treatments. However, the large amount of data generated by microarrays requires effective reduction of discriminant gene features into reliable sets of tumor biomarkers for such multiclass tumor discrimination. The availability of reliable sets of biomarkers, especially serum biomarkers, should have a major impact on our understanding and treatment of cancer. RESULTS We have combined genetic algorithm (GA) and all paired (AP) support vector machine (SVM) methods for multiclass cancer categorization. Predictive features can be automatically determined through iterative GA/SVM, leading to very compact sets of non-redundant cancer-relevant genes with the best classification performance reported to date. Interestingly, these different classifier sets harbor only modest overlapping gene features but have similar levels of accuracy in leave-one-out cross-validations (LOOCV). Further characterization of these optimal tumor discriminant features, including the use of nearest shrunken centroids (NSC), analysis of annotations and literature text mining, reveals previously unappreciated tumor subclasses and a series of genes that could be used as cancer biomarkers. With this approach, we believe that microarray-based multiclass molecular analysis can be an effective tool for cancer biomarker discovery and subsequent molecular cancer diagnosis.

FTO gene polymorphisms and obesity risk: a meta-analysis.

BackgroundThe pathogenesis of obesity is reportedly related to variations in the fat mass and an obesity-associated gene (FTO); however, as the number of reports increases, particularly with respect to varying ethnicities, there is a need to determine more precisely the effect sizes in each ethnic group. In addition, some reports have claimed ethnic-specific associations with alternative SNPs, and to that end there has been a degree of confusion.MethodsWe searched PubMed, MEDLINE, Web of Science, EMBASE, and BIOSIS Preview to identify studies investigating the associations between the five polymorphisms and obesity risk. Individual study odds ratios (OR) and their 95% confidence intervals (CI) were estimated using per-allele comparison. Summary ORs were estimated using a random effects model.ResultsWe identified 59 eligible case-control studies in 27 articles, investigating 41,734 obesity cases and 69,837 healthy controls. Significant associations were detected between obesity risk and the five polymorphisms: rs9939609 (OR: 1.31, 95% CI: 1.26 to 1.36), rs1421085 (OR: 1.43, 95% CI: 1.33 to 1.53), rs8050136 (OR: 1.25, 95% CI: 1.13 to 1.38), rs17817449 (OR: 1.54, 95% CI: 1.41 to 1.68), and rs1121980 (OR: 1.34, 95% CI: 1.10 to 1.62). Beggs and Eggers tests provided no evidence of publication bias for the polymorphisms except rs1121980. There is evidence of higher heterogeneity, with I2 test values ranging from 38.1% to 84.5%.ConclusionsThis meta-analysis suggests that FTO may represent a low-penetrance susceptible gene for obesity risk. Individual studies with large sample size are needed to further evaluate the associations between the polymorphisms and obesity risk in various ethnic populations.

The association risk of male subfertility and testicular cancer: a systematic review.

Background An association between male subfertility and an increased risk of testicular cancer has been proposed, but conflicting results of research on this topic have rendered this theory equivocal. To more precisely assess the association between subfertility and the risk of testicular cancer, we performed a systematic review of international epidemiologic evidence. Principal Findings We searched the Medline database for records from January 1966 to March 2008 complemented with manual searches of the literature and then identified studies that met our inclusion criteria. Study design, sample size, exposure to subfertility and risk estimates of testicular cancer incidence were abstracted. Summary relative risks (RRs) with 95% confidence intervals (CIs) were calculated using the DerSimonian and Laird model. All statistical tests were two-sided. We identified seven case-control studies and two cohort studies published between 1987 and 2005. Analysis of the seven case-control studies that included 4,954 participants revealed an overall statistically significant association between subfertility and increased risk of testicular cancer (summary RR = 1.68, 95% CI: 1.22 to 2.31), without heterogeneity between studies (Q = 8.46, P heterogeneity = 0.21, I 2 statistics = 0.29). The association between subfertility and testicular cancer was somewhat stronger in the United States (summary RR = 1.75, 95% CI: 1.01 to 3.02) than it was in Europe (summary RR = 1.53, 95% CI: 1.22 to 1.92). The source of the control subjects had a statistically significant effect on the magnitude of the association (population-based summary—RR = 2.15, 95% CI: 1.11 to 4.17; hospital-based summary—RR = 1.56, 95% CI: 0.93 to 2.61). After excluding possible cryptorchidism, an important confounding factor, we also found a positive association between subfertility and increased risk of testicular cancer (summary RR = 1.59, 95% CI: 1.28 to 1.98). These results were consistent between studies conducted in the United States and in Europe (Q = 0.20, P heterogeneity = 0.66). Of the two cohort studies that reported standardized incidence ratios, both reported a statistically significant positive association between subfertility and increased risk of testicular cancer. Conclusions Our findings support a relationship between subfertility and increased risk of testicular cancer and apply to the management of men with subfertility, and prevention and diagnosis of testicular cancer.

TCF7L2 gene polymorphisms and type 2 diabetes risk: a comprehensive and updated meta-analysis involving 121 174 subjects

Recently, many new loci associated with type 2 diabetes have been uncovered by genetic association studies and genome-wide association studies. As more reports are made, particularly with respect to varying ethnicities, there is a need to determine more precisely the effect sizes in each major racial group. In addition, some reports have claimed ethnic-specific associations with alternative single-nucleotide polymorphisms (SNPs), and to that end there has been a degree of confusion. We conducted a meta-analysis using an additive genetic model. Eight polymorphisms in 155 studies with 121174 subjects (53385 cases and 67789 controls) were addressed in this meta-analysis. Significant associations were found between type 2 diabetes and rs7903146, rs12255372, rs11196205, rs7901695, rs7895340 and rs4506565, with summary odds ratios (ORs) (95% confidence interval) of 1.39 (1.34-1.45), 1.33 (1.27-1.40), 1.20 (1.14-1.26), 1.32 (1.25-1.39), 1.21 (1.13-1.29) and 1.39 (1.29-1.49), respectively. In addition, no significant associations were found between the two polymorphisms (rs290487 and rs11196218) and type 2 diabetes. The summary ORs for the six statistically significant associations (P < 0.05) were further evaluated by estimating the false-positive report probability, with results indicating that all of the six significant associations were considered noteworthy, and may plausibly be true associations. Significant associations were found between the six polymorphisms (rs7903146, rs12255372, rs11196205, rs7901695, rs7895340 and rs4506565) in the TCF7L2 gene and type 2 diabetes risk, and the other two polymorphisms (rs11196218 and rs290487) were not found to be significantly associated with type 2 diabetes. Subgroups analyses show that significant associations are not found between the six SNPs (rs7903146, rs12255372, rs11196205, rs7901695, rs7895340, and rs4506565) and the type 2 diabetes in some ethnic populations.

Multi-class cancer classification through gene expression profiles: microRNA versus mRNA.

Both microRNA (miRNA) and mRNA expression profiles are important methods for cancer type classification. A comparative study of their classification performance will be helpful in choosing the means of classification. Here we evaluated the classification performance of miRNA and mRNA profiles using a new data mining approach based on a novel SVM (Support Vector Machines) based recursive feature elimination (nRFE) algorithm. Computational experiments showed that information encoded in miRNAs is not sufficient to classify cancers; gut-derived samples cluster more accurately when using mRNA expression profiles compared with using miRNA profiles; and poorly differentiated tumors (PDT) could be classified by mRNA expression profiles at the accuracy of 100% versus 93.8% when using miRNA profiles. Furthermore, we showed that mRNA expression profiles have higher capacity in normal tissue classifications than miRNA. We concluded that classification performance using mRNA profiles is superior to that of miRNA profiles in multiple-class cancer classifications.

Characterization of microRNAs in cephalochordates reveals a correlation between microRNA repertoire homology and morphological similarity in chordate evolution

SUMMARY Cephalochordates, urochordates, and vertebrates comprise the three extant groups of chordates. Although higher morphological and developmental similarity exists between cephalochordates and vertebrates, molecular phylogeny studies have instead suggested that the morphologically simplified urochordates are the closest relatives to vertebrates. MicroRNAs (miRNAs) are regarded as the major factors driving the increase of morphological complexity in early vertebrate evolution, and are extensively characterized in vertebrates and in a few species of urochordates. However, the comprehensive set of miRNAs in the basal chordates, namely the cephalochordates, remains undetermined. Through extensive sequencing of a small RNA library and genomic homology searches, we characterized 100 miRNAs from the cephalochordate amphioxus, Branchiostoma japonicum, and B. floridae. Analysis of the evolutionary history of the cephalochordate miRNAs showed that cephalochordates possess 54 miRNA families homologous to those of vertebrates, which is threefold higher than those shared between urochordates and vertebrates. The miRNA contents demonstrated a clear correlation between the extent of miRNA overlapping and morphological similarity among the three chordate groups, providing a strong evidence of miRNAs being the major genetic factors driving morphological complexity in early chordate evolution.

SVM–T-RFE: A novel gene selection algorithm for identifying metastasis-related genes in colorectal cancer using gene expression profiles

Although metastasis is the principal cause of death cause for colorectal cancer (CRC) patients, the molecular mechanisms underlying CRC metastasis are still not fully understood. In an attempt to identify metastasis-related genes in CRC, we obtained gene expression profiles of 55 early stage primary CRCs, 56 late stage primary CRCs, and 34 metastatic CRCs from the expression project in Oncology (http://www.intgen.org/expo/). We developed a novel gene selection algorithm (SVM-T-RFE), which extends support vector machine recursive feature elimination (SVM-RFE) algorithm by incorporating T-statistic. We achieved highest classification accuracy (100%) with smaller gene subsets (10 and 6, respectively), when classifying between early and late stage primary CRCs, as well as between metastatic CRCs and late stage primary CRCs. We also compared the performance of SVM-T-RFE and SVM-RFE gene selection algorithms on another large-scale CRC dataset and the five public microarray datasets. SVM-T-RFE bestowed SVM-RFE algorithm in identifying more differentially expressed genes, and achieving highest prediction accuracy using equal or smaller number of selected genes. A fraction of selected genes have been reported to be associated with CRC development or metastasis.

−8p12–23 and +20q Are Predictors of Subtypes and Metastatic Pathways in Colorectal Cancer: Construction of Tree Models Using Comparative Genomic Hybridization Data

A substantial body of evidence suggests the genetic heterogeneous pattern and multiple pathways in colorectal cancer initiation and progression. In this study, we construct a branching tree and multiple distance-based tree models to elucidate these genetic patterns and pathways in colorectal cancer by using a data set comprised of 244 cases of comparative genomic hybridization. We identify the six most common gains of chromosomal regions of 7p (37.0%), 7q11-32 (34.8%), 8q (48.3%), 13q (49.1%), 20p (36.1%), and 20q (50.4%), and the nine most common losses of 1p13-36 (30.9%), 4p15 (24.3%), 4q33-34 (24.3%), 8p12-23 (50.9%), 15q13-14 (23.5%), 15q24-25 (24.3%), 17p (34.8%), 18p (36.5%), and 18q (61.7%) in colorectal cancer. We classify colorectal cancer into two distinct groups: one preceding with -8p12-23, and the other with +20q. The sample-based classification tree also demonstrates that colorectal cancer can be classified into multiple subtypes marked by -8p12-23 and +20q. By comparing chromosomal abnormalities between primary and metastatic colorectal cancer, we identify five potential metastatic pathways: (-18q, -18p), (-8p12-23, -4p15, -4q33-34), (+20q, +20p), (+20q, +7p, +7q11-32), and +8q. -8p12-23 and +20q are inferred to be the two marker events of colorectal cancer metastasis. The current oncogenetic tree models may contribute to our understanding towards molecular genetics in colorectal cancer. Particularly, the metastatic pathways we describe may provide pivotal clues for metastatic candidate genes, and thus impact on the prediction and intervention of metastatic colorectal cancer.

A New Implementation of Recursive Feature Elimination Algorithm for Gene Selection from Microarray Data

We proposed a new approach for gene selection and multi-cancer classification based on step-by-step improvement of classification performance (SSiCP).The SSiCP gene selection algorithms were evaluated over the NCI60 and GCM benchmark datasets, with an accuracy of 96.6% and 95.5% in 10-fold cross validation,respectively. Furthermore, the SSiCP outperformed recently published algorithms when applied to another two multi-cancer data sets.Computational evidence indicated that SSiCP can avoid over fitting effectively. Compared with various gene selection algorithms, the implementation of SSiCPis very simple, and all the computational experiments are repeatable.