Lien-Chin Chen

Identification of tyrosine-phosphorylated proteins associated with lung cancer metastasis using label-free quantitative analyses.

Lung cancer is a lethal disease, and early metastasis is the major cause of treatment failure and cancer-related death. Tyrosine phosphorylated (P-Tyr) proteins are involved in the invasive and metastatic behavior of lung cancer; however, only a limited number of targets were identified. We attempt to characterize P-Tyr proteins and events involved in the metastatic process. In a previous work, we have developed a strategy for identification of protein phosphorylation. Here, this strategy was used to characterize the tyrosine phosphoproteome of lung cancer cells that have different invasive abilities (CL1-0 vs. CL1-5). Using our analytical strategy, we report the identification of 335 P-Tyr sites from 276 phosphoproteins. Label-free quantitative analysis revealed that 36 P-Tyr peptides showed altered levels between CL1-0 and CL1-5 cells. From this list of sites, we extracted two novel consensus sequences and four known motifs for specific kinases and phosphatases including EGFR, Src, JAK2, and TC-PTP. Protein-protein interaction network analysis of the altered P-Tyr proteins illustrated that 11 proteins were linked to a network containing EGFR, c-Src, c-Myc, and STAT, which is known to be related to lung cancer metastasis. Among these 11 proteins, 7 P-Tyr proteins have not been previously reported to be associated with lung cancer metastasis and are of greatest interest for further study. The characterized tyrosine phosphoproteome and altered P-Tyr targets may provide a better comprehensive understanding of the mechanisms of lung cancer invasion/metastasis and discover potential therapies.

Combining Alkaline Phosphatase Treatment and Hybrid Linear Ion Trap/Orbitrap High Mass Accuracy Liquid Chromatography−Mass Spectrometry Data for the Efficient and Confident Identification of Protein Phosphorylation

Protein phosphorylation is a vital post-translational modification that is involved in a variety of biological processes. Several mass spectrometry-based methods have been developed for phosphoprotein characterization. In our previous work, we demonstrated the capability of a computational algorithm in mining phosphopeptide signals in large LC-MS data sets by measuring the mass shifts due to phosphatase treatment (Wu, H. Y.; Tseng, V. S.; Liao, P. C. J. Proteome Res. 2007, 6, 1812-1821). Mass accuracy seems to play an important role in efficiently selecting out phosphopeptide signals. In recent years, the hybrid linear ion trap (LTQ)/Orbitrap mass spectrometer, which provides a high mass accuracy, has emerged as a powerful instrument in proteomic analysis. Here, we developed a process to incorporate LC-MS data that was generated from an LTQ/Orbitrap mass spectrometer into our strategy for taking advantage of the accurate mass measurement. LTQ/Orbitrap raw files were converted to the open file format mzXML via the ReAdW.exe program. To find peaks that were contained in each mzXML file, an open-source computer program, msInspect, was utilized to locate isotopes and assemble those isotopes into peptides. An in-house program, LcmsFormatConverter, was utilized for signal filtering and format transformation. A proposed in-house program, DeltaFinder, was modified and used for defining signals with an exact mass shift due to the dephosphorylation reaction, which generated a table that listed potential phosphopeptide signals. The retention times and m/z values of these selected LC-MS signals were used to program subsequent LC-MS/MS experiments to get high-confidence phosphorylation site determination. Compared to our previous work finished by using a quadrupole/time-of-flight mass spectrometer, a larger number of phosphopeptides in the casein mixture were identified by using LTQ/Orbitrap data, demonstrating the merit of high mass accuracy in our strategy. In addition, the characterization of the lung cancer cell tyrosine phosphoproteome revealed that the use of alkaline phosphatase treatment combined with accurate mass measurement in this strategy increased data repeatability and confidence.

A Statistical Procedure to Selectively Detect Metabolite Signals in LC-MS Data Based on Using Variable Isotope Ratios

The tracing of metabolite signals in LC-MS data using stable isotope-labeled compounds has been described in the literature. However, the filtering efficiency and confidence when mining metabolite signals in complex LC-MS datasets can be improved. Here, we propose an additional statistical procedure to increase the compound-derived signal mining efficiency. This method also provides a highly confident approach to screen out metabolite signals because the correlation of varying concentration ratios of native/stable isotope-labeled compounds and their instrumental response ratio is used. An in-house computational program [signal mining algorithm with isotope tracing (SMAIT)] was developed to perform the statistical procedure. To illustrate the SMAIT concept and its effectiveness for mining metabolite signals in LC-MS data, the plasticizer, di-(2-ethylhexyl) phthalate (DEHP), was used as an example. The statistical procedure effectively filtered 15 probable metabolite signals from 3617 peaks in the LC-MS data. These probable metabolite signals were considered structurally related to DEHP. Results obtained here suggest that the statistical procedure could be used to confidently facilitate the detection of probable metabolites from a compound-derived precursor presented in a complex LC-MS dataset.

Malignant pleural effusion cells show aberrant glucose metabolism gene expression

Malignant pleural effusion (MPE) accompanying lung adenocarcinoma indicates poor prognosis and early metastasis. This study aimed to identify genes related to MPE formation. Three tissue sample cohorts, seven from healthy lungs, 18 from stage I–III lung adenocarcinoma with adjacent healthy lung tissue and 13 from lung adenocarcinomas with MPE, were analysed by oligonucleotide microarray. The identified genes were verified by quantitative real-time PCR (qRT-PCR), immunohistochemical staining, and immunofluorescence confocal microscopy. 20 up- or down-regulated genes with a two-fold change in MPE cancer cells compared to healthy tissues were differentially expressed from early- to late-stage lung cancer. Of 13 genes related to cellular metabolism, aldolase A (ALDOA), sorbitol dehydrogenase (SORD), transketolase (TKT), and tuberous sclerosis 1 (TSC1) were related to glucose metabolism. qRT-PCR validated their mRNA expressions in pleural metastatic samples. Immunohistochemical staining confirmed aberrant TKT, ALDOA, and TSC1 expressions in tumour cells. Immunofluorescence confirmed TKT co-localisation and co-distribution of ALDOA with thyroid transcription factor 1-positive cancer cells. TKT regulated the proliferation, vascular endothelial growth factor secretion in vitro and in vivo vascular permeability of cancer cell. Glucose metabolic reprogramming by ALDOA, SORD, TKT and TSC1 is important in MPE pathogenesis.

A multi-objective genetic-fuzzy mining algorithm

In this paper, we propose a multi-objective genetic-fuzzy mining algorithm for extracting both membership functions and association rules from quantitative transactions. Two objective functions are used to find the Pareto front. The first one is the suitability of membership functions. It consists of two factors, coverage factor and overlap factor, to avoid two bad types of membership functions. The second one is the total number of large 1-itemsets from a given set of minimum support values. The two criteria have a trade-off relationship. Experimental results also show the effectiveness of the proposed approach in finding the Pareto-front membership functions.

Gene Relation Discovery by Mining Similar Subsequences in Time-Series Microarray Data

Time-series microarray techniques are newly used to monitor large-scale gene expression profiles for studying biological systems. Previous studies have discovered novel regulatory relations among genes by analyzing time-series microarray data. In this study, we investigate the problem of mining similar subsequences in time-series microarray data so as to discover novel gene relations. A functional relationship among genes often presents itself by locally similar and potentially time-shifted patterns in their expression profiles. Although a number of studies have been done on time-series data analysis, they are insufficient in handling four important issues for time-series microarray data analysis, namely scaling, offset, shift, and noise. We proposed a novel method to address the four issues simultaneously, which consists of three phase, namely angular transformation, symbolic transformation and suffix-tree-based similar subsequences searching. Through experimental evaluation, it is shown that our method can effectively discover biological relations among genes by identifying the similar subsequences. Moreover, the execution efficiency of our method is much better than other approaches

Hybrid data mining approaches for prevention of drug dispensing errors

Prevention of drug dispensing errors is an importance topic in medical care. In this paper, we propose a risk management approach, namely Hybrid Data Mining (HDM), to prevent the problem of drug dispensing errors. An intelligent drug dispensing errors prevention system based on the proposed approach is then implemented. The proposed approach consists of two main procedures: First, the classification modeling and logistic regression approaches are used to derive decision tree and regression function from the given dispensing errors cases and drug databases. In the second procedure, similar drugs are then gathered together into clusters by combing clustering technique (PoCluster) and the extracted logistic regression function. The drugs that may cause dispensing errors will then be alerted through the clustering results and the decision tree. Through experimental evaluation on real datasets in a medical center, the proposed approach was shown to be capable of discovering the potential dispensing errors effectively. Hence, the proposed approach and implemented system serve as very useful application of data mining techniques for risk management in healthcare fields.

A two-phase clustering approach for peak alignment in mining mass spectrometry data

In recent years, the mass spectrometry technologies emerge as useful tools for biomarker discovery through studying protein profiles in various biological specimens. In mining mass spectrometry datasets, peak alignment is a critical issue among the preprocessing steps that affect the quality of analysis results. In this paper, we proposed a novel algorithm named Two-Phases Clustering for peak Alignment (TPC-Align) to align mass spectrometry peaks across samples in the pre-processing phase. The TPC-Align algorithm sequentially considers the distribution of intensity values and the locations of mass-to-charge ratio values of peaks between samples. Moreover, TPC-Align algorithm can also report a list of significantly differential peaks between samples, which serve as the candidate biomarkers for further biological study. The proposed peak alignment method was compared to the current peak alignment approach based on one-dimension hierarchical clustering through experimental evaluations, and the results show that TPC-Align outperforms the traditional method on the real dataset.

Constrained clustering for gene expression data mining

Constrained clustering algorithms have the advantage that domaindependent constraints can be incorporated in clustering so as to achieve better clustering results. However, the existing constrained clustering algorithms are mostly k-means like methods, which may only deal with distance-based similarity measures. In this paper, we propose a constrained hierarchical clustering method, called Correlational-Constrained Complete Link (C-CCL), for gene expression analysis with the consideration of gene-pair constraints, while using correlation coefficients as the similarity measure. C-CCL was evaluated for the performance with the correlational version of COP-k-Means (C-CKM) method on a real yeast dataset. We evaluate both clustering methods with two validation measures and the results show that C-CCL outperforms C-CKM substantially in clustering quality

A Cluster-Based Divide-and-Conquer Genetic-Fuzzy Mining Approach for Items with Multiple Minimum Supports

In this paper, an enhanced efficient approach for speeding up the evolution process for finding minimum supports, membership functions and fuzzy association rules is proposed by utilizing clustering techniques. All the chromosomes use the requirement satisfaction derived only from the representative chromosomes in the clusters and from their own suitability of membership functions to calculate the fitness values. The evaluation cost can thus be greatly reduced due to the cluster-based time-saving process. The final best minimum supports and membership functions in all the populations are then gathered together for mining fuzzy association rules. Experimental results also show the efficiency of the proposed approach.