Whole exome sequencing and MicroRNA profiling of lung adenocarcinoma identified risk prediction features for tumors at stage I and its substages.

OBJECTIVES: About 20% of stage I lung adenocarcinoma (LUAD) patients suffer a relapse after surgical resection. While finer substages have been defined and refined in the AJCC staging system, clinical investigations on the tumor molecular landscape are lacking. MATERIALS AND METHODS: We performed whole exome sequencing, DNA copy number and microRNA profiling on paired tumor-normal samples from a cohort of 113 treatment-naïve stage I Taiwanese LUAD patients. We searched for molecular features associated with relapse-free survival (RFS) of stage I or its substages and validated the findings with an independent Caucasian LUAD cohort. RESULTS: We found sixteen nonsynonymous mutations harbored at EGFR, KRAS, TP53, CTNNB1 and six other genes associated with poor RFS in a dose-dependent manner via variant allele fraction (VAF). An index, maxVAF, was constructed to quantify the overall mutation load from genes other than EGFR. High maxVAF scores discriminated a small group of high-risk LUAD at stage I (median RFS: 4.5 versus 69.5 months; HR = 10.5, 95% CI = 4.22-26.12, P < 0.001). At the substage level, higher risk was found for patients with high maxVAF or high miR-31; IA (median RFS: 32.1 versus 122.8 months, P = 0.005) and IB (median RFS: 7.1 versus 26.2, P = 0.049). MicroRNAs, miR-182, miR-183 and miR-196a were found correlated with EGFR mutation and poor RFS in stage IB patients. CONCLUSION: Distinctive features of somatic gene mutation and microRNA expression of stage I LUAD are characterized to complement the survival prognosis by substaging. The findings open up more options for precision management of stage I LUAD patients.

Feature selection translates drug response predictors from cell lines to patients.

Targeted therapies and chemotherapies are prevalent in cancer treatment. Identification of predictive markers to stratify cancer patients who will respond to these therapies remains challenging because patient drug response data are limited. As large amounts of drug response data have been generated by cell lines, methods to efficiently translate cell-line-trained predictors to human tumors will be useful in clinical practice. Here, we propose versatile feature selection procedures that can be combined with any classifier. For demonstration, we combined the feature selection procedures with a (linear) logit model and a (non-linear) K-nearest neighbor and trained these on cell lines to result in LogitDA and KNNDA, respectively. We show that LogitDA/KNNDA significantly outperforms existing methods, e.g., a logistic model and a deep learning method trained by thousands of genes, in prediction AUC (0.70-1.00 for seven of the ten drugs tested) and is interpretable. This may be due to the fact that sample sizes are often limited in the area of drug response prediction. We further derive a novel adjustment on the prediction cutoff for LogitDA to yield a prediction accuracy of 0.70-0.93 for seven drugs, including erlotinib and cetuximab, whose pathways relevant to anti-cancer therapies are also uncovered. These results indicate that our methods can efficiently translate cell-line-trained predictors into tumors.

Genetic Disruption of KLF1 K74 SUMOylation in Hematopoietic System Promotes Healthy Longevity in Mice.

The quest for rejuvenation and prolonged lifespan through transfusion of young blood has been studied for decades with the hope of unlocking the mystery of the key substance(s) that exists in the circulating blood of juvenile organisms. However, a pivotal mediator has yet been identified. Here, atypical findings are presented that are observed in a knockin mouse model carrying a lysine to arginine substitution at residue 74 of Krüppel-like factor 1 (KLF1/EKLF), the SUMOylation-deficient Klf1K74R/K74R mouse, that displayed significant improvement in geriatric disorders and lifespan extension. Klf1K74R/K74R mice exhibit a marked delay in age-related physical performance decline and disease progression as evidenced by physiological and pathological examinations. Furthermore, the KLF1(K74R) knockin affects a subset of lymphoid lineage cells; the abundance of tumor infiltrating effector CD8+ T cells and NKT cells is increased resulting in antitumor immune enhancement in response to tumor cell administration. Significantly, infusion of hematopoietic stem cells (HSCs) from Klf1K74R/K74R mice extends the lifespan of the wild-type mice. The Klf1K74R/K74R mice appear to be an ideal animal model system for further understanding of the molecular/cellular basis of aging and development of new strategies for antiaging and prevention/treatment of age-related diseases thus extending the healthspan as well as lifespan.

A Positive Regulatory Feedback Loop between EKLF/KLF1 and TAL1/SCL Sustaining the Erythropoiesis.

The erythroid Krüppel-like factor EKLF/KLF1 is a hematopoietic transcription factor binding to the CACCC DNA motif and participating in the regulation of erythroid differentiation. With combined use of microarray-based gene expression profiling and the promoter-based ChIP-chip assay of E14.5 fetal liver cells from wild type (WT) and EKLF-knockout (Eklf-/-) mouse embryos, we identified the pathways and direct target genes activated or repressed by EKLF. This genome-wide study together with the molecular/cellular analysis of the mouse erythroleukemic cells (MEL) indicate that among the downstream direct target genes of EKLF is Tal1/Scl. Tal1/Scl encodes another DNA-binding hematopoietic transcription factor TAL1/SCL, known to be an Eklf activator and essential for definitive erythroid differentiation. Further identification of the authentic Tal gene promoter in combination with the in vivo genomic footprinting approach and DNA reporter assay demonstrate that EKLF activates the Tal gene through binding to a specific CACCC motif located in its promoter. These data establish the existence of a previously unknow positive regulatory feedback loop between two DNA-binding hematopoietic transcription factors, which sustains mammalian erythropoiesis.

Landscape of Mitochondria Genome and Clinical Outcomes in Stage 1 Lung Adenocarcinoma.

Risk factors including genetic effects are still being investigated in lung adenocarcinoma (LUAD). Mitochondria play an important role in controlling imperative cellular parameters, and anomalies in mitochondrial function might be crucial for cancer development. The mitochondrial genomic aberrations found in lung adenocarcinoma and their associations with cancer development and progression are not yet clearly characterized. Here, we identified a spectrum of mitochondrial genome mutations in early-stage lung adenocarcinoma and explored their association with prognosis and clinical outcomes. Next-generation sequencing was used to reveal the mitochondrial genomes of tumor and conditionally normal adjacent tissues from 61 Stage 1 LUADs. Mitochondrial somatic mutations and clinical outcomes including relapse-free survival (RFS) were analyzed. Patients with somatic mutations in the D-loop region had longer RFS (adjusted hazard ratio, adjHR = 0.18, p = 0.027), whereas somatic mutations in mitochondrial Complex IV and Complex V genes were associated with shorter RFS (adjHR = 3.69, p = 0.012, and adjHR = 6.63, p = 0.002, respectively). The risk scores derived from mitochondrial somatic mutations were predictive of RFS (adjHR = 9.10, 95%CI: 2.93-28.32, p < 0.001). Our findings demonstrated the vulnerability of the mitochondrial genome to mutations and the potential prediction ability of somatic mutations. This research may contribute to improving molecular guidance for patient treatment in precision medicine.

Tumor mutation burden and recurrent tumors in hereditary lung cancer.

Lung cancer is the leading cause of cancer death worldwide and cancer relapse accounts for the majority of cancer mortality. The mechanism is still unknown, especially in hereditary lung cancer without known actionable mutations. To identify genetic alternations involved in hereditary lung cancer and relapse is urgently needed. We collected genetic materials from a unique hereditary lung cancer patient's blood, first cancer tissue (T1), adjacent normal tissue (N1), relapse cancer tissue (T2), and adjacent normal tissue (N2) for whole genome sequencing. We identified specific mutations in T1 and T2, and attributed them to tumorigenesis and recurrence. These tumor specific variants were enriched in antigen presentation pathway. In addition, a lung adenocarcinoma cohort from the TCGA dataset was used to confirm our findings. Patients with high mutation burdens in tumor specific genes had decreased relapse-free survival (P = 0.017, n = 186). Our study may provide important insight for designing immunotherapeutic treatment for hereditary lung cancer.

Association of Bone Metastasis With Early-Stage Breast Cancer in Women With and Without Precancer Osteoporosis According to Osteoporosis Therapy Status.

Importance: Deaths from cancer are attributed more to secondary than primary tumors, but the pathogenesis of organ-specific cancer metastasis has not been determined. Objective: To investigate whether precancer osteoporosis and osteoporosis therapy are associated with alteration of bone metastasis patterns. Design, Setting, and Participants: This nationwide retrospective cohort study was performed from January 1, 2002, to December 31, 2013, using 2 cohorts from the Taiwan National Health Insurance Research Database: a random sample of 1 million beneficiaries from the Longitudinal Health Insurance Database who were enrolled in 2005 (LHID2005) and a specific data set of all the patients with osteoporosis. Patients diagnosed with breast cancer and precancer osteoporosis from January 1, 2002, to December 31, 2011, were included in the study, and their records were examined through December 31, 2013. From LHID2005, we selected 9104 women with early-stage breast cancer, of whom 705 had precancer osteoporosis. We identified 29 183 patients from the cohort of patients with breast cancer and osteoporosis, 14 020 of whom had precancer osteoporosis. Data analysis was performed from December 31, 2016, to August 31, 2018. Exposures: Precancer osteoporosis and osteoporosis therapy. Main Outcomes and Measures: The risk of bone metastasis in patients with and without precancer osteoporosis and patients receiving and not receiving osteoporosis therapy as well as time to bone metastasis development. Results: Among 9104 patients with breast cancer from the Longitudinal Health Insurance Database (mean [SD] age, 46.7 [14.0] years), precancer osteoporosis was not associated with a difference in risk of bone metastasis (adjusted hazard ratio [aHR], 0.87; 95% CI, 0.58-1.30; P = .49). Among 14 020 patients with precancer osteoporosis from the other cohort (mean [SD] age, 58.9 [11.6] years), osteoporosis therapy had no association with the risk of bone metastasis (bisphosphonates: aHR, 1.47; 95% CI, 1.00-2.17; P = .05; nonbisphosphonate drugs: aHR, 1.00; 95% CI, 0.72-1.39; P > .99). Compared with those without precancer osteoporosis (median time to bone metastasis development, 2.87 years; interquartile range [IQR], 1.34-4.86 years), among those with precancer osteoporosis, the median time to develop bone metastasis was shorter in those who did not receive treatment (1.74 years; IQR, 0.58-3.60 years; P < .001), whereas this time was the same for those who received treatment (bisphosphonates: 2.34 years; IQR, 1.23-3.13 years; nonbisphosphonate drugs: 2.08 years; IQR, 0.92-4.95 years). Conclusions and Relevance: Precancer osteoporosis was not associated with risk of bone metastasis, but untreated osteoporosis was associated with accelerated progression of bone metastasis when it occurred. Organ microenvironments interact with disseminated cancer mostly after the specific organ has been predetermined to be the designated location. Because recurrences and metastases are major obstacles to cancer treatments, determining which organs may be at risk for metastases may be crucial to treating the disease.

Serum Glutamine Levels as a Potential Diagnostic Biomarker in Sepsis following Surgery for Peritonitis.

Few diagnostic biomarkers for sepsis after emergency peritonitis surgery are available to clinicians, and, thus, it is important to develop new biomarkers for patients undergoing this procedure. We investigated whether serum glutamine and selenium levels could be diagnostic biomarkers of sepsis in individuals recovering from emergency peritonitis surgery. From February 2012 to March 2013, patients who had peritonitis diagnosed at the emergency department and underwent emergency surgery were screened for eligibility. Serum glutamine and selenium levels were obtained at pre-operative, post-operative and recovery time points. The average level of pre-operation serum glutamine was significantly different from that on the recovery day (0.317 ± 0.168 vs. 0.532 ± 0.155 mM, P < 0.001); moreover, serum glutamine levels were unaffected by surgery. Selenium levels were significantly lower on the day of surgery than they were at recovery (106.6 ± 36.39 vs. 130.68 ± 56.98 ng/mL, P = 0.013); no significant difference was found between pre-operation and recovery selenium levels. Unlike selenium, glutamine could be a sepsis biomarker for individuals with peritonitis. We recommend including glutamine as a biomarker for sepsis severity assessment in addition to the commonly used clinical indicators.

R331W Missense Mutation of Oncogene YAP1 Is a Germline Risk Allele for Lung Adenocarcinoma With Medical Actionability.

PURPOSE: Adenocarcinoma is the most dominant type of lung cancer in never-smoker patients. The risk alleles from genome-wide association studies have small odds ratios and unclear biologic roles. Here we have taken an approach featuring suitable medical actionability to identify alleles with low population frequency but high disease-causing potential. PATIENTS AND METHODS: Whole-genome sequencing was performed for a family with an unusually high density of lung adenocarcinoma with available DNA from the affected mother, four affected daughters, and one nonaffected son. Candidate risk alleles were confirmed by matrix-assisted laser desorption ionization time of flight mass spectroscopy. Validation was conducted in an external cohort of 1,135 participants without cancer and 1,312 patients with lung adenocarcinoma. Family follow-ups were performed by genotyping the relatives of the original proband and the relatives of the identified risk-allele carriers. Low-dose computed tomography scans of the chest were evaluated for lung abnormalities. RESULTS: YAP1 R331W missense mutation from the original family was identified and validated in the external controls and the cohort with lung adenocarcinoma. The YAP1 mutant-allele carrier frequency was 1.1% in patients with lung adenocarcinoma compared with 0.18% in controls (P = .0095), yielding an odds ratio (adjusted for age, sex, and smoking status) of 5.9. Among the relatives, YAP1-mutant carriers have overwhelmingly higher frequencies of developing lung adenocarcinoma or ground-glass opacity lung lesions than those who do not carry the mutation (10:0 v 1:7; P < .001). YAP1 mutation was shown to increase the colony formation ability and invasion potential of lung cancer cells. CONCLUSION: These results implicated YAP1 R331W as an allele predisposed for lung adenocarcinoma with high familial penetrance. Low-dose computed tomography scans may be recommended to this subpopulation, which is at high risk for lung cancer, for personalized prevention and health management.

Shisa3 is associated with prolonged survival through promoting ß-catenin degradation in lung cancer.

RATIONALE: Despite advances in treatment and prognosis of non-small cell lung cancer (NSCLC), patient outcomes are still unsatisfactory. OBJECTIVES: To reduce the morbidity and mortality of patients with NSCLC, a more comprehensive understanding of mechanisms involved in cancer progression is urgently needed. METHODS: By comparison of gene expression profiles in the cell line pair with differential invasion ability, CL1-0 and CL1-5, we found that Shisa3 was highly expressed in the low invasive cells. The effect of Shisa3 on invasion, migration, proliferation, apoptosis, epithelial-mesenchymal transition, and anchorage-independent growth activities in vitro and on tumor growth and metastasis in mice models were examined. The underlying mechanism of Shisa3 was explored by microarray and pathway analysis. Finally, the correlation of Shisa3 expression and clinical outcome was also calculated. MEASUREMENTS AND MAIN RESULTS: We identified Shisa3 as a novel tumor suppressor, which induces ß-catenin degradation resulting in suppression of tumorigenesis and invasion in vitro. Shisa3 decreased the tumor growth in mice with subcutaneous implantation and reduced the number of metastatic nodules in mice with tail vein injection and orthotopic implantation. Shisa3 performs the tumor suppression activity through WNT signaling predicted by microarray analysis. Our data found that Shisa3 accelerates ß-catenin degradation and was positively associated with overall survival and progression-free survival of NSCLC. CONCLUSIONS: Our results reveal that Shisa3 acts as a tumor suppressor by acceleration of ß-catenin degradation and provide new insight for cancer prognosis and therapy.

Cancer-associated fibroblasts regulate the plasticity of lung cancer stemness via paracrine signalling.

Cancer stem cells (CSCs) are a promising target for treating cancer, yet how CSC plasticity is maintained in vivo is unclear and is difficult to study in vitro. Here we establish a sustainable primary culture of Oct3/4(+)/Nanog(+) lung CSCs fed with CD90(+) cancer-associated fibroblasts (CAFs) to further advance our knowledge of preserving stem cells in the tumour microenvironment. Using transcriptomics we identify the paracrine network by which CAFs enrich CSCs through de-differentiation and reacquisition of stem cell-like properties. Specifically, we find that IGF1R signalling activation in cancer cells in the presence of CAFs expressing IGF-II can induce Nanog expression and promote stemness. Moreover, this paracrine signalling predicts overall and relapse-free survival in stage I non-small cell lung cancer (NSCLC) patients. IGF-II/IGF1R signalling blockade inhibits Nanog expression and attenuates cancer stem cell features. Our data demonstrate that CAFs constitute a supporting niche for cancer stemness, and targeting this paracrine signalling may present a new therapeutic strategy for NSCLC.

Focal amplification of HOXD-harboring chromosome region is implicated in multiple-walled carbon nanotubes-induced carcinogenicity.

Multiple-walled carbon nanotubes (MWCNTs) may cause carcinogenesis. We found that long-term exposure to MWCNTs can induce irreversible oncogenic transformation of human bronchial epithelial cells and tumorigenicity in vivo. A genome-wide array-comparative genomic hybridization (aCGH) analysis revealed global chromosomal aberration in MWCNTs-treated clones, predominantly at chromosome 2q31-32, where the potential oncogenes HOXD9 and HOXD13 are located. Functional assays confirmed that this variation can modulate oncogenic signaling and plays a part in MWCNTs-induced tumorigenesis, suggesting that MWCNTs are carcinogens that act by altering genomic stability and oncogenic copy numbers.

Genome-wide analysis of three-way interplay among gene expression, cancer cell invasion and anti-cancer compound sensitivity.

BACKGROUND: Chemosensitivity and tumor metastasis are two primary issues in cancer management. Cancer cells often exhibit a wide range of sensitivity to anti-cancer compounds. To gain insight on the genetic mechanism of drug sensitivity, one powerful approach is to employ the panel of 60 human cancer cell lines developed by the National Cancer Institute (NCI). Cancer cells also show a broad range of invasion ability. However, a genome-wide portrait on the contributing molecular factors to invasion heterogeneity is lacking. METHODS: Our lab performed an invasion assay on the NCI-60 panel. We identified invasion-associated (IA) genes by correlating our invasion profiling data with the Affymetrix gene expression data on NCI-60. We then employed the recently released chemosensitivity data of 99 anti-cancer drugs of known mechanism to investigate the gene-drug correlation, focusing on the IA genes. Afterwards, we collected data from four independent drug-testing experiments to validate our findings on compound response prediction. Finally, we obtained published clinical and molecular data from two recent adjuvant chemotherapy cohorts, one on lung cancer and one on breast cancer, to test the performance of our gene signature for patient outcome prediction. RESULTS: First, we found 633 IA genes from the invasion-gene expression correlation study. Then, for each of the 99 drugs, we obtained a subset of IA genes whose expression levels correlated with drug-sensitivity profiles. We identified a set of eight genes (EGFR, ITGA3, MYLK, RAI14, AHNAK, GLS, IL32 and NNMT) showing significant gene-drug correlation with paclitaxel, docetaxel, erlotinib, everolimus and dasatinib. This eight-gene signature (derived from NCI-60) for chemosensitivity prediction was validated by a total of 107 independent drug tests on 78 tumor cell lines, most of which were outside of the NCI-60 panel. The eight-gene signature predicted relapse-free survival for the lung and breast cancer patients (log-rank P = 0.0263; 0.00021). Multivariate Cox regression yielded a hazard ratio of our signature of 5.33 (95% CI = 1.76 to 16.1) and 1.81 (95% CI = 1.19 to 2.76) respectively. The eight-gene signature features the cancer hallmark epidermal growth factor receptor (EGFR) and genes involved in cell adhesion, migration, invasion, tumor growth and progression. CONCLUSIONS: Our study sheds light on the intricate three-way interplay among gene expression, invasion and compound-sensitivity. We report the finding of a unique signature that predicts chemotherapy survival for both lung and breast cancer. Augmenting the NCI-60 model with in vitro characterization of important phenotype-like invasion potential is a cost-effective approach to power the genomic chemosensitivity analysis.

Tissue-specific gene expression templates for accurate molecular characterization of the normal physiological states of multiple human tissues with implication in development and cancer studies.

BACKGROUND: To elucidate the molecular complications in many complex diseases, we argue for the priority to construct a model representing the normal physiological state of a cell/tissue. RESULTS: By analyzing three independent microarray datasets on normal human tissues, we established a quantitative molecular model GET, which consists of 24 tissue-specific Gene Expression Templates constructed from a set of 56 genes, for predicting 24 distinct tissue types under disease-free condition. 99.2% correctness was reached when a large-scale validation was performed on 61 new datasets to test the tissue-prediction power of GET. Network analysis based on molecular interactions suggests a potential role of these 56 genes in tissue differentiation and carcinogenesis.Applying GET to transcriptomic datasets produced from tissue development studies the results correlated well with developmental stages. Cancerous tissues and cell lines yielded significantly lower correlation with GET than the normal tissues. GET distinguished melanoma from normal skin tissue or benign skin tumor with 96% sensitivity and 89% specificity. CONCLUSIONS: These results strongly suggest that a normal tissue or cell may uphold its normal functioning and morphology by maintaining specific chemical stoichiometry among genes. The state of stoichiometry can be depicted by a compact set of representative genes such as the 56 genes obtained here. A significant deviation from normal stoichiometry may result in malfunction or abnormal growth of the cells.

Clustered genomic alterations in chromosome 7p dictate outcomes and targeted treatment responses of lung adenocarcinoma with EGFR-activating mutations.

PURPOSE: Although epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) have been proven more effective for patients with lung adenocarcinoma with EGFR-activating mutation rather than wild type, the former group still includes approximately 30% nonresponders. The molecular basis of this substantial response heterogeneity is unknown. Our purpose was to seek molecular aberrations contributing to disease progression at the genome-wide level and identify the prognostic signature unique to patients with EGFR-activating mutation. PATIENTS AND METHODS: We first investigated the molecular differences between tumors with EGFR-activating mutation and wild-type tumors by conducting high-density array comparative genomic hybridization on a collection of 138 adenocarcinoma tissues. We then used an independent group of 114 patients to validate the clinical relevance of copy-number alterations (CNAs) in predicting overall and disease-free survival. Finally, focusing on 23 patients with EGFR mutation receiving EGFR-TKI treatment, we investigated the association between CNAs and response to EGFR-TKIs. RESULTS: We identified chromosome regions with differential CNAs between tumors with EGFR-activating mutation and wild-type tumors and found the aberration sites to cluster highly on chromosome 7p. A cluster of six representative chromosome 7p genes predicted overall and disease-free survival for patients with EGFR-activating mutation but not for those with wild type. Importantly, simultaneous presence of more genes with increased CNAs in this cluster correlated with less favorable response to EGFR-TKIs in patients with EGFR-activating mutation. CONCLUSION: Our results shed light on why responses to EGFR-TKIs are heterogeneous among patients with EGFR-activating mutation. They may lead to better patient management in this population.

A four-gene signature from NCI-60 cell line for survival prediction in non-small cell lung cancer.

PURPOSE: Metastasis is the main cause of mortality in non-small cell lung cancer (NSCLC) patients. Genes that can discriminate the invasion ability of cancer cells may become useful candidates for clinical outcome prediction. We identify invasion-associated genes through computational and laboratorial approach that supported this idea in NSCLC. EXPERIMENTAL DESIGN: We first conducted invasion assay to characterize the invasion abilities of NCI-60 lung cancer cell lines. We then systematically exploited NCI-60 microarray databases to identify invasion-associated genes that showed differential expression between the high and the low invasion cell line groups. Furthermore, using the microarray data of Duke lung cancer cohort (GSE 3141), invasion-associated genes with good survival prediction potentials were obtained. Finally, we validated the findings by conducting quantitative PCR assay on an in-house collected patient group (n = 69) and by using microarray data from two public western cohorts (n = 257 and 186). RESULTS: The invasion-associated four-gene signature (ANKRD49, LPHN1, RABAC1, and EGLN2) had significant prediction in three validation cohorts (P = 0.0184, 0.002, and 0.017, log-rank test). Moreover, we showed that four-gene signature was an independent prognostic factor (hazard ratio, 2.354, 1.480, and 1.670; P = 0.028, 0.014, and 0.033), independent of other clinical covariates, such as age, gender, and stage. CONCLUSION: The invasion-associated four-gene signature derived from NCI-60 lung cancer cell lines had good survival prediction power for NSCLC patients.

A method for analyzing censored survival phenotype with gene expression data.

BACKGROUND: Survival time is an important clinical trait for many disease studies. Previous works have shown certain relationship between patients' gene expression profiles and survival time. However, due to the censoring effects of survival time and the high dimensionality of gene expression data, effective and unbiased selection of a gene expression signature to predict survival probabilities requires further study. METHOD: We propose a method for an integrated study of survival time and gene expression. This method can be summarized as a two-step procedure: in the first step, a moderate number of genes are pre-selected using correlation or liquid association (LA). Imputation and transformation methods are employed for the correlation/LA calculation. In the second step, the dimension of the predictors is further reduced using the modified sliced inverse regression for censored data (censorSIR). RESULTS: The new method is tested via both simulated and real data. For the real data application, we employed a set of 295 breast cancer patients and found a linear combination of 22 gene expression profiles that are significantly correlated with patients' survival rate. CONCLUSION: By an appropriate combination of feature selection and dimension reduction, we find a method of identifying gene expression signatures which is effective for survival prediction.

A system for enhancing genome-wide coexpression dynamics study.

Statistical similarity analysis has been instrumental in elucidation of the voluminous microarray data. Genes with correlated expression profiles tend to be functionally associated. However, the majority of functionally associated genes turn out to be uncorrelated. One conceivable reason is that the expression of a gene can be sensitively dependent on the often-varying cellular state. The intrinsic state change has to be plastically accommodated by gene-regulatory mechanisms. To capture such dynamic coexpression between genes, a concept termed "liquid association" (LA) has been introduced recently. LA offers a scoring system to guide a genome-wide search for critical cellular players that may interfere with the coexpression of a pair of genes, thereby weakening their overall correlation. Although the LA method works in many cases, a direct extension to more than two genes is hindered by the "curse of dimensionality." Here we introduce a strategy of finding an informative 2D projection to generalize LA for multiple genes. A web site is constructed that performs on-line LA computation for any user-specified group of genes. We apply this scoring system to study yeast protein complexes by using the Saccharomyces cerevisiae protein complexes database of the Munich Information Center for Protein Sequences. Human genes are also investigated by profiling of 60 cancer cell lines of the National Cancer Institute. In particular, our system links the expression of the Alzheimer's disease hallmark gene APP (amyloid-beta precursor protein) to the beta-site-cleaving enzymes BACE and BACE2, the gamma-site-cleaving enzymes presenilin 1 and 2, apolipoprotein E, and other Alzheimer's disease-related genes.