Prognostic gene expression signature associated with two molecularly distinct subtypes of colorectal cancer

Patients and gene expression data
All clinical and gene expression data are available from the
National Center for Biotechnology Information Gene Expression
Omnibus database (http://www.ncbi.nlm.nih.gov/geo). Gene
expression data from the Moffit Cancer Center (Moffit cohort,
GSE17536, n¼177) were used as the exploration data set.11 Gene
expression data from the Vanderbilt Medical Center (GSE17537,
n¼55) and Max Planck Institute (GSE12945, n¼62) were pooled
and used as the first validation data set (Vanderbilt and Max
Planck (VMP) cohort, n¼117).11 12 Gene expression data from
the Royal Melbourne Hospital that is part of GSE14333 (n¼96)
were used as the second validation data set.13 Gene expression
data of these patients were redeposited as an independent data
set to Gene Expression Omnibus (GSE29971). To test the prognostic
significance of gene expression signatures, we used only
gene expression data with available patient survival data.
Although three prognostic variables (OS, disease-specific survival
and disease-free survival (DFS)) were available for the Moffit
cohort, only OS and DFS data were available for the VMP and
Melbourne cohorts, respectively.
Adjuvant chemotherapy data were available only for the
Moffit, Vanderbilt Medical Center and Melbourne cohorts. Of
the 328 patients in the Moffit, Vanderbilt and Melbourne
cohorts, 147 (2 in AJCC stage I, 28 in stage II, 81 in stage III and
36 in stage IV) had received standard adjuvant chemotherapy
(either single-treatment 5-fluorouracil/capecitabine or a combination
of 5-fluorouracil and oxaliplatin). The remaining patients
did not receive chemotherapy (n¼168) or treatment data were
not available (n¼13).

Disease free survival - DFS

In order to analyze survival time and disease-free survival time, the
following variables are needed:

Dfsevent has the value 1 if the patient died or relapsed, 0 otherwise.

Dfstime is the time, in months, from surgery to death or relapse if either occurred.
Otherwise, it is the time that the patient was observed after surgery.

Survevent has the value 1 if the patient died, 0 otherwise.

Survtime is the time, in months, from surgery to death if the patient died. Otherwise, it is the
time that the patient was observed after surgery.



support.sas.com/publishing/pubcat/chaps/58416.pdf
support.sas.com/publishing/pubcat/chaps/55504.pdf

or in dropbox


In FDA:
http://www.google.com.sg/url?sa=t&rct=j&q=definition%20of%20disease-free%20survival&source=web&cd=8&cad=rja&ved=0CEwQFjAH&url=http%3A%2F%2Fwww.fda.gov%2Fdownloads%2FDrugs%2FGuidanceComplianceRegulatoryInformation%2FGuidances%2Fucm071590.pdf&ei=DY08UMzjJY-3rAeahIG4BA&usg=AFQjCNEe0Mi0RyMYfk8POmoLrJl0lS1r2w&sig2=pOqOro76BC_HG9mI4Xz9mQ

http://tor.imng.com/tor/breast/70918.html

Bevacizumab-capecitabine may favor ER-positive women

Adding the antiangiogenesis agent bevacizumab to capecitabine prolonged time to progression for 106 women who received the combination as first-line therapy for metastatic breast cancer in a multicenter phase II trial.

The advantage was small, but an unplanned subset analysis surprised investigators with the observation that estrogen receptor-positive (ER+) women were about twice as likely to benefit as were those with estrogen receptor-negative (ER-) disease.

At a median follow-up of 12.9 months, overall time to progression was 5.7 months in the initial intent-to-treat analysis. The interval stretched to 8.9 months for 57 ER+ women in the subset analysis, but was only 4 months for 49 ER- women.

Although median overall survival had not yet been reached at the time of the report, it was projected to be more than 16 months for the group as a whole and more than 16.6 months for ER+ women, but only 7.5 months for ER- women.

The objective response rate based on complete and partial responses followed a similar pattern: 38% for the full cohort, 47% for ER+ women, but were 27% for those who were ER-. The differences were statistically significant (P = .0001).

At the time that the study was presented, 10 ER+ women, but none of the ER- women, were still receiving first-line treatment.

"This study suggests the combination is more active in estrogen receptor-positive patients than in estrogen receptor-negative patients, albeit with the caveat that this is an unplanned subset analysis," the lead author, Dr. George W. Sledge, told the audience.

Dr. Sledge, Ballve Professor of Medicine and Pathology at Indiana University, Indianapolis, said the researchers found the disparity when they tried to figure out why the combination did not yield a longer time to progression. The trial barely met its primary end point of 5.6 months, which was better than the 4 months reported for capecitabine (Xeloda) monotherapy at the time the trial was designed. Nonetheless, it was not as long as had been anticipated, and Dr. Sledge characterized the results overall as "somewhat disappointing."

Dr. Stephen R.D. Johnston agreed in a discussion of the study that the findings merited further investigation. "The end point was met by a matter of 3 days, but ... the difference in estrogen receptors is clearly hypothesis generating and of interest," said Dr. Johnston of the Royal Marsden NHS Foundation Trust in Chelsea, England. "The better effect seen in estrogen receptor-positive patients is hitherto unexplained."

Hoffmann-La Roche, maker of capecitabine, sponsored the trial.

Most of the women in the trial were white, with a median age of 56.8 years (range, 36-82 years). All had HER2-negative disease. Although most had received prior neoadjuvant or adjuvant therapy, none had had antiangiogenic or fluoropyrimidine therapy, and none had received adjuvant treatment within the previous 6 months.

Patients received 1,000 mg/m² of capecitabine twice daily for the first 2 weeks of a 3-week cycle and 15 mg/kg of intravenous bevacizumab (Avastin) on the first day of each cycle. Treatment continued until progression, at which point patients were to start second-line therapy with bevacizumab.

Dr. Sledge described the combination as well tolerated, with toxicity in line with expectations for capecitabine (diarrhea, stomatitis, and hand-foot-and-mouth syndrome) and bevacizumab (hypertension). Still to come are reports on analyses of serum samples and on second-line therapy in patients who progressed.

"Analysis of serum levels of VEGF [vascular endothelial growth factor] will be crucial to our understanding of what we have observed," Dr. Johnston commented, alluding to bevacizumab's mechanism of action as an inhibitor of VEGF. "It is not beyond the realm of possibility that VEGF levels are higher in ER-positive women," he said.

Sledge G. et al. Safety and efficacy of capecitabine (C) plus bevacizumab (B) as first-line in metastatic breast cancer. Abstract 1013.

Commentary

The study by Sledge et al. follows a randomized trial that compared capecitabine to capecitabine and bevacizumab in patients who had received prior therapy for metastatic disease. This earlier trial suggested bevacizumab added little benefit in this population. The subsequent Eastern Cooperative Oncology Group (ECOG) E2100 study in patients with no prior therapy for metastatic disease demonstrated, however, that the addition of bevacizumab to paclitaxel resulted in a higher response rate and progression-free survival than did paclitaxel alone. As a result XCALBR was designed to readdress the efficacy of bevacizumab and capecitabine. The results reported by Sledge et al. suggest the combination of capecitabine and bevacizumab results in a modest improvement in the expected response rate and progression-free survival over what might be expected with single-agent capecitabine as reported in previous studies. The most intriguing aspect of the study is the apparent marked increase in response rate and progression-free survival for the subset of patients with estrogen receptor-positive disease. There is no ready explanation for this observation, but it will be studied further with laboratory correlative studies yet to be reported. — William J. Gradishar, M.D.

Invasive??

Notch-dependent VEGFR3 upregulation allows angiogenesis without VEGF-VEGFR2 signaling (Nature, March 2012). Featured Review How do miRNAs Regulate Angiogenesis? How does Extracellular Matrix (ECM) Induce Angiogenesis? miRNAs Regulate Angiogenesis
Angiogenesis and Tumor Metastasis
Angiogenesis Angiogenic Growth Factors & Inhibitors Hypoxia ECM and Cell Adhesion	EMT (Epithelial-Mesenchymal Transition) Tumor Metastasis Notch Signaling Focal Adhesions

Example

###################################################
#
data <- exprs (GCCL.37$eset)# eset of GCCL.37
#1. repeated centering
data.centering <- medpolish(data)$residual

#2. repeated standardization
std.data <- data.centering
for (i in 1:10)
{
#By row (gene)
std.data <- t(scale(t(std.data), scale=T))
#By column (array)
std.data <- scale(std.data, scale=T)
}

HC <- hclust(dist(t(std.data)))
plot (HC) # Hclust plot of 2 large natural clusters of 37 GCCL


cl.37 <- cutree(HC, k=2) # 2 large clusters 16 G1 and 21 G2
setwd("E:\\Projects\\8.ComBAT\\ComBat201\\G1G2")
write.table(cl.37, file="G1G2.cl.37.medpolish.txt", sep="\t")


cl.37 <- cutree(HC, k=3) # 2 large clusters 16 G1 and 21 G2
setwd("E:\\Projects\\8.ComBAT\\ComBat201\\G1G2")
write.table(cl.37, file="G1G2.cl.37.medpolish3.txt", sep="\t")

HC

http://smd.stanford.edu/gp/pages/protocols/ClassDiscovery_hier.html

Hierarchical Clustering

protocols

Cluster genes and/or samples based on how close they are to one another. The result is a tree structure, referred to as dendrogram.

Before you begin

Gene expression data must be in a GCT or RES file.
Example file: all_aml_test.gct.

learn more: file formats

Step 1: PreprocessDataset

Preprocess gene expression data to remove platform noise and genes that have little variation. Although researchers generally preprocess data before clustering if doing so removes relevant biological information, skip this step.

• Open module in the GenePattern window.
• Open module with example data in the GenePattern window.

CONSIDERATIONS

• PreprocessDataset can preprocess the data in one or more ways (in this order):
  1. Set threshold and ceiling values. Any value lower/higer than the threshold/ceiling value is reset to the threshold/ceiling value.
  2. Convert each expression value to the log base 2 of the value.
  3. Remove genes (rows) if a given number of its sample values are less than a given threshold.
  4. Remove genes (rows) that do not have a minimum fold change or expression variation.
  5. Discretize or normalize the data.
• When using ratios to compare gene expression between samples, convert values to log base 2 of the value to bring up- and down-regulated genes to the same scale. For example, ratios of 2 and .5 indicating two-fold changes for up- and down-regulated expression, respectively, are converted to +1 and -1.
• If you did not generate the expression data, check whether preprocessing steps have already been taken before running the PreprocessDataset module.

learn more: PreprocessDataset

Step 2: HierarchicalClustering

Run hierarchical clustering on genes and/or samples to create dendrograms for the clustered genes (*.gtr) and/or clustered samples (*.atr), as well as a file (*.cdt) that contains the original gene expression data ordered to reflect the clustering.

• Open HierarchicalClustering
• Open HierarchicalClustering with example data

CONSIDERATIONS

• Best practice is to normalize (row/column normalize parameters) and center (row/column center parameters) the data being clustered.
• The CDT output file must be converted to a GCT file before it can be used as an input file for another GenePattern module (other than HierachicalClusteringViewer). For instructions on converting a CDT file to a GCT file, see Creating Input Files.

learn more: HierarchicalClustering

Step 3: HierarchicalClusteringViewer

Display a heat map of the clustered gene expression data, with dendrograms showing how the genes and/or samples were clustered.

• Open HierarchicalClusteringViewer
• Open HierarchicalClusteringViewer with example data

CONSIDERATIONS

• Select File>Save Image to save the heat map and dendrograms to an image file. Supported formats include bmp, eps, jpeg, png, and tiff.

learn more: HierarchicalClusteringViewer

http://smd.stanford.edu/gp/pages/protocols/ClassDiscovery_consensus.html

Best practice is to normalize the data being clustered
ZL: usually done by standardizing on row (gene) then on column (array),  or 

Adjust Cycle 1) log transform ##### skip this step if RMA

Adjust Cycle 2) median center genes and arrays 

repeat (2) five to ten times  #### like median polish 

Adjust Cycle 3) normalize genes and arrays

repeat (3) five to ten times

see cluster3 (http://db.tt/fKAluEip treeview) documentation.



http://bonsai.hgc.jp/~mdehoon/software/cluster/manual/Data.html#Data

Consensus Clustering

protocols

Determine an optimal number of clusters by repeatedly running a selected clustering algorithm. Examine the resulting consensus matrix to assess the stability of the resulting clusters.

Before you begin

Gene expression data must be in a GCT or RES file.
Example file: all_aml_test.gct.

learn more: file formats

Step 1: PreprocessDataset

Preprocess gene expression data to remove platform noise and genes that have little variation. Although researchers generally preprocess data before clustering if doing so removes relevant biological information, skip this step.

• Open module in the GenePattern window.
• Open module with example data in the GenePattern window.

CONSIDERATIONS

• PreprocessDataset can preprocess the data in one or more ways (in this order):
  1. Set threshold and ceiling values. Any value lower/higer than the threshold/ceiling value is reset to the threshold/ceiling value.
  2. Convert each expression value to the log base 2 of the value.
  3. Remove genes (rows) if a given number of its sample values are less than a given threshold.
  4. Remove genes (rows) that do not have a minimum fold change or expression variation.
  5. Discretize or normalize the data.
• When using ratios to compare gene expression between samples, convert values to log base 2 of the value to bring up- and down-regulated genes to the same scale. For example, ratios of 2 and .5 indicating two-fold changes for up- and down-regulated expression, respectively, are converted to +1 and -1.
• If you did not generate the expression data, check whether preprocessing steps have already been taken before running the PreprocessDataset module.

learn more: PreprocessDataset

Step 2: ConsensusClustering

ConsensusClustering runs a selected clustering algorithm (by default, hierarchical clustering) against perturbations of the gene expression data a selected number of times (by default, 20). It assesses the stability of the resulting clusters by creating a consensus matrix.

For every pair of objects, the matrix records the number of times both are assigned to the same cluster divided by the number of times both are in the perturbed data set. A consensus matrix where all values are 0 or 1 corresponds to perfect consensus.

• Open ConsensusClustering
• Open ConsensusClustering with example data

CONSIDERATIONS

• ConsensusClustering clusters genes or samples, not both.
• ConsensusClustering groups objects (genes or samples) into k clusters. It groups objects into two clusters, then three clusters, up to the maximum number of clusters specified by the kmaxparameter (by default, 5). The module builds a separate consensus matrix for each set of clusters.
• Best practice is to normalize the data being clustered (normalize type parameter).

learn more: ConsensusClustering

Step 3: HeatMapViewer

Run the HeatMapViewer module to view the consensus matrices. The consensus matrix is formatted as a GCT file. The HeatMapViewer displays the consensus matrix as a heat map. A consensus matrix where all values are dark blue (0) or dark red (1) corresponds to perfect consensus.

• Open HeatMapViewer
• Open HeatMapViewer with example data

CONSIDERATIONS

• For more about the consensus matrix and its interpretation, see Monti et al., 2003.
• ConsensusClustering also creates a text file (*.clu) listing the items belonging to each cluster, text files (*.clsdist, *.stdev) listing the cluster statistics, and a .pdf file showing statistical plots (Lorenz curve, Gini index, Consensus CDF) that can be used to determine the best number of clusters. To display any of these files, click the file.

share internet between two laptops

http://hi.baidu.com/clark_kent/blog/item/4b11962f9fdc3e311e308928.html

http://hairongmao.blog.163.com/blog/static/90258012010149336267/

http://www.360doc.com/content/09/0304/11/108458_2706190.shtml

The key must be 5 digits as shown below.

CSC and hypermethylation

http://www.ncbi.nlm.nih.gov/pubmed/18820729
http://genome.cshlp.org/content/early/2012/03/05/gr.131169.111

icgc dataset

ftp://data.dcc.icgc.org/version_8/Pancreatic_Cancer-QCMG-AU/

TARGET data
http://target.nci.nih.gov/dataMatrix/TARGET_DataMatrix.html

MAS5

QC
rm(list=ls())
library(affy)
library(simpleaffy)
setwd("/home/leiz/SG201")

data <- ReadAffy()
data <- ReadAffy()

data.mas5 <- mas5(data)
## Or you can try mas5
genes <- log2(as.matrix(exprs(data.mas5 )))
write.table(genes, file="SG201.mas5.txt", sep="\t")

gastric cancer

http://www.cancer.net/patient/Cancer+Types/Stomach+Cancer/ci.Stomach+Cancer.printer

Statistics

This year, an estimated 21,320 adults (13,020 men and 8,300 women) in the United States will be diagnosed with stomach cancer. It is estimated that 10,540 deaths (6,190 men and 4,350 women) from this disease will occur this year.

The incidence of stomach cancer varies in different parts of the world. Although it is decreasing in the Western world, it is still one of the most common cancer types worldwide.

The five-year survival rate (percentage of people who survive at least five years after the cancer is detected, excluding those who die from other diseases) of people with stomach cancer is about 26%. This statistic reflects the fact that most people with stomach cancer are diagnosed after the cancer has already spread to other parts of the body. If stomach cancer is found before it has spread, the five-year survival rate is generally higher but depends on the stage of the cancer found during surgery.

Cancer survival statistics should be interpreted with caution. These estimates are based on data from thousands of people with this type of cancer in the United States each year, but the actual risk for a particular individual may differ. It is not possible to tell a person how long he or she will live with stomach cancer. Because the survival statistics are measured in five-year intervals, they may not represent advances made in the treatment or diagnosis of this cancer. Learn more about understanding statistics.

Statistics adapted from the American Cancer Society's publication, Cancer Facts & Figures 2012.

http://www.cancernetwork.com/cancer-management/gastric/article/10165/1802600

Survival

Most patients still present with advanced disease, and their survival remains poor. From 1999 to 2006, only 23% of patients with gastric cancer presented with localized disease. The relative 5-year survival rate for gastric cancer of all stages is 26%.

NGS training @GIS - continued

###########day 2 de novo assemble


PATH=/mnt/software/assemblers/velvet:/usr/local/samtools:/usr/local/samtools/bcftools:/mnt/software/mappers/bwa:$PATH

#1. create simulated data
cd ~/refgenome
/mnt/software/mappers/samtools/samtools-0.1.7/misc/wgsim reference.fa NC_1.fastq NC_2.fastq
cd ~/prj2
mv ../refgenome/NC_*.fastq ./
shuffleSequences_fastq.pl ./NC_1.fastq ./NC_2.fastq NC.fastq

#2. velvet
velveth velvet 21 -fastq -shortPaired NC.fastq
velvetg velvet -ins_length 300 -exp_cov 1 -read_trkg yes -amos_file yes



#############day 2 ANNOVAR
PATH=$PATH:/home/javeda/ANNOVAR/annovar
cd ~/prj3
cp ../prj1/var.filt.vcf ./
convert2annovar.pl ./var.filt.vcf -format vcf4 -includeinfo >var.filt.vcf.avinput
awk 'BEGIN {FS="\t"} {$1="chr17"; $2=$2+46000000; $3=$3+46000000; print}' var.filt.vcf.avinput >variant1.avinput

#Gene-based annotation
annotate_variation.pl -buildver hg19 variant1.avinput /home/javeda/ANNOVAR/annovar/humandb/
more variant1.avinput.exonic_variant_function

#Region-based annotation
annotate_variation.pl -dbtype band -regionanno  -buildver hg19 variant1.avinput /home/javeda/ANNOVAR/annovar/humandb/
more variant1.avinput.hg19_cytoBand

#Filter-based annotation
annotate_variation.pl -buildver hg19 -filter -dbtype snp131 variant1.avinput /home/javeda/ANNOVAR/annovar/humandb/
more variant1.avinput.hg19_cytoBand

#snps found in dbsnp131
more variant1.avinput.hg19_snp131_dropped

#snps not in dbsnp131
more variant1.avinput.hg19_snp131_filtered

NGS training @GIS

PATH=/usr/local/samtools:/usr/local/samtools/bcftools:/mnt/software/mappers/bwa:$PATH

put files to ~/refgenome/reference.fa ~/prj1/reads.fastq

#1. index step
cd ~/refgenome
bwa index -a is reference.fa
samtools faidx ./reference.fa

#2. align and pileup
cd ~/prj1
bwa aln ../refgenome/reference.fa ./reads.fastq >reads.bwa.aln.sai
bwa samse ../refgenome/reference.fa ./reads.bwa.aln.sai reads.fastq >reads.bwa.aln.sam
samtools import ../refgenome/reference.fa.fai ./reads.bwa.aln.sam reads.bwa.aln.bam
samtools sort reads.bwa.aln.bam reads.bwa.aln.sorted
samtools index reads.bwa.aln.sorted.bam
samtools mpileup -f ../refgenome/reference.fa reads.bwa.aln.sorted.bam >temp.pileup
samtools mpileup -ugf ../refgenome/reference.fa reads.bwa.aln.sorted.bam >reads.bwa.aln.sorted.pileup

#3. call variant file
bcftools view -bcvg reads.bwa.aln.sorted.pileup >var.row.bcf
bcftools view var.row.bcf |vcfutils.pl varFilter -D 100 >var.filt.vcf
more var.filt.vcf

#4. download and install tablet
http://bioinf.scri.ac.uk/tablet/download.shtml
use tablet to view reads.bwa.aln.sorted.bam

Batch Effect in AU

date2col <- function(date.list)
{
 clr.template = c("red", "orange", "yellow", "green", "cyan", "blue", "purple")
 num.dates <- length(date.list)
 clr.list <- vector()
 clr.list[1] <- "red"
 c.index <- 0
 for (i in 2:num.dates) {
  if(date.list[i] == date.list[i-1]) {
  clr.list[i] = clr.list[i-1]
  } else {
  c.index <- c.index+1
  clr.list[i] = clr.template[c.index %% 7+1]
  }
 }
 return(clr.list)
}


wk.dir <- "E:\\CEL\\GastricCancer\\AU\\PM_data_new\\Gastric_Affy_files\\Tumors"
setwd(wk.dir)

file.info.file <- "files.info.user.batch.txt"
file.info <- read.table(file=file.info.file, header=T, row.names=1)
file.info[order(file.info$EXP_DATE),]
file.info <- file.info[order(as.POSIXct(strptime(file.info$EXP_DATE, "%m/%d/%Y"))),]
my.color <- file.info$EXP_DATE
my.color <- date2col(my.color)


data <- read.table(file="AU_GC70.rma.txt", header=T, row.names=1)
data.ctrl <- data[54614:54675, rownames(file.info)]
library("gplots")



data <- sweep(data.ctrl, 1, apply(data.ctrl, 1, median)) #just median centered
data[data < -4] <- -4
data[data > 4] <- 4
hm<-heatmap.2(as.matrix(data), col=greenred(75), scale="none", dendrogram="none", Rowv= T, Colv=F, ColSideColors=my.color, key=TRUE, symkey=FALSE, density.info="none",trace="none", cexRow=0.75,cexCol=0.5)
pdf(file = "Batch_in_CtrlGenes1.pdf", width=10, height=10)
hm<-heatmap.2(as.matrix(data), col=greenred(75), scale="none", dendrogram="none", Rowv= T, Colv=F, ColSideColors=my.color, key=TRUE, symkey=FALSE, density.info="none",trace="none", cexRow=0.75,cexCol=0.5)
dev.off()


data <- t(scale(t(data.ctrl), scale=T)) #standardized by row(gene)
data[data < -3] <- -3
data[data > 3] <- 3
hm<-heatmap.2(as.matrix(data), col=greenred(75), scale="none", dendrogram="none", Rowv= T, Colv=F, ColSideColors=my.color, key=TRUE, symkey=FALSE, density.info="none",trace="none", cexRow=0.75,cexCol=0.5)
pdf(file = "Batch_in_CtrlGenes2.pdf", width=10, height=10)
hm<-heatmap.2(as.matrix(data), col=greenred(75), scale="none", dendrogram="none", Rowv= T, Colv=F, ColSideColors=my.color, key=TRUE, symkey=FALSE, density.info="none",trace="none", cexRow=0.75,cexCol=0.5)
dev.off()



genes<-data.ctrl
genes<-t(genes)
pcs<-prcomp(genes)
summary(pcs) #select first N=10 PCs depending on Cumulative Proportion (e.g. >= 97.7%)
#pcs$x[,1:10]
#write.table(pcs$x[,1:15], file=ctrl.genes.pcs, sep = "\t")
#pcs<-prcomp(data[1:22215,])
library(scatterplot3d)
PC1<-pcs$x[,1]
PC2<-pcs$x[,2]
PC3<-pcs$x[,3]


group.colors <- my.color
group.colors <- file.info$COLOR



scatterplot3d(PC1,PC2,PC3, main="PCA scatterplot before ComBat normalization", color=group.colors, pch=16)

scatterplot3d(PC3,PC2,PC1, main="PCA scatterplot before ComBat normalization", color=group.colors, pch=16)

scatterplot3d(PC2,PC1,PC3, main="PCA scatterplot before ComBat normalization", color=group.colors, pch=16)

Proliferative

http://www.drugbank.ca/drugs/DB01229

harmacodynamics

Paclitaxel is a taxoid antineoplastic agent indicated as first-line and subsequent therapy for the treatment of advanced carcinoma of the ovary, and other various cancers including breast cancer. Paclitaxel is a novel antimicrotubule agent that promotes the assembly of microtubules from tubulin dimers and stabilizes microtubules by preventing depolymerization. This stability results in the inhibition of the normal dynamic reorganization of the microtubule network that is essential for vital interphase and mitotic cellular functions. In addition, paclitaxel induces abnormal arrays or "bundles" of microtubules throughout the cell cycle and multiple asters of microtubules during mitosis.

Mechanism of action

Paclitaxel interferes with the normal function of microtubule growth. Whereas drugs like colchicine cause the depolymerization of microtubules in vivo, paclitaxel arrests their function by having the opposite effect; it hyper-stabilizes their structure. This destroys the cell's ability to use its cytoskeleton in a flexible manner. Specifically, paclitaxel binds to the β subunit of tubulin. Tubulin is the "building block" of mictotubules, and the binding of paclitaxel locks these building blocks in place. The resulting microtubule/paclitaxel complex does not have the ability to disassemble. This adversely affects cell function because the shortening and lengthening of microtubules (termed dynamic instability) is necessary for their function as a transportation highway for the cell. Chromosomes, for example, rely upon this property of microtubules during mitosis. Further research has indicated that paclitaxel induces programmed cell death (apoptosis) in cancer cells by binding to an apoptosis stopping protein called Bcl-2 (B-cell leukemia 2) and thus arresting its function.

http://annonc.oxfordjournals.org/content/18/suppl_12/xii15.short
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2361361/
http://www.genecards.org/cgi-bin/carddisp.pl?gene=MAPT

http://www.springerlink.com/content/m68011x7147r7051/

Resistance to taxanes, related to limited efficacy of systemic therapy in cancer patients, is multifactorial. Among mechanisms of resistance to taxanes, those related to microtubule-associated proteins (MAP), including protein Tau, are of great importance. Protein Tau (50–64 kD) binds to beta-tubulin in the same place as paclitaxel. In preclinical studies, low expression of Tau in cancer cells was associated with increased sensitivity to paclitaxel. High expression of Tau protein in ER-positive breast cancers indicates resistance to taxane-containing chemotherapy and sensitivity to hormonal treatment. This article reviews current knowledge on predictive value of protein Tau in response to taxanes. Better understanding of its role may facilitate patients selection to this sort of treatment and lead to treatment optimization.

YGC gastric cancer

http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE13861

Test of propotions

inv <- c(54, 13) pro <- c(82, 29) meta <- c(56, 14) patients <- c(192, 56) prop.test(inv, patients) prop.test(pro, patients) prop.test(meta, patients)

silhouette width

library(cluster)
setwd("E:\\Projects\\8.ComBAT\\ComBat248")
rmafile <- "ComBat248T_filtered.stdized.txt"
data <- read.table(rmafile, header = T, sep = "\t", row.names=1)

setwd("E:\\Projects\\8.ComBAT\\ComBat248\\CCP_FIS_SG248_top10000_ZL20120301\\Run3")
feature.file <- "K3_limma_geneset4.txt"
feature <- read.table(feature.file, header = T, sep = "\t", row.names=1)

data <- data[feature$ID,]
plot (hclust(dist(t(data)))) # Hclust plot of data

dmat <- dist(t(data))
hc <- hclust(dmat, "average")
hsil <- silhouette(cutree(hc, k=3), dmat)
plot(hsil)
write.table(hsil, "silhouette_width.avg_linkage.txt", sep="\t")