Extracting phenotype type by genotype using GenABEL

GenABEL is an excellent R package for GWAS studies. It uses special data structure to efficiently store data. The data structure is quite useful and results in remarkable time saving when running GWAS, it does have some limitations. For example, often in GWAS studies one need to know phenotype distribution across genotype of some variant but I couldn’t find a straightforward way of looking at phenotype distribution across genotypes (there may be a better way of doing it but I couldn’t find it)

To get phenotypic information across genotypes I used the following approach (assuming that data is in an object called ‘data’

1. Abstract phenotypic information
pheno<- phdata(data)
The returned object is a dataframe and can be confirmed with class(pheno) command

2. Abstract SNP data
snps <- as.character(data[, c("SNP1", "SNP2", "SNP3")])
You can change as.character in the line above with as.numeric if you want to get genotype information in 0,1,2 format.
The returned object is a matrix with row numbers as subject ID. Thus we need to do two things with this matrix. First we have to convert it into a dataframe and then we have to convert rownames into a column of id

3. Convert matrix 'snps' into a dataframe with row names as an additional column
snps.df<-data.frame(as.numeric(rownames(snps)),snps)
colnames(snps.df)[1]="id"                          ### Change the column name to 'id'
snp.data <- merge(pheno, snps.df, by="id")  

Now you have a dataframe with phenotype data and SNPs genotype data

Downloading and Merging NHANES datasets in R

The National Health and Nutrition Examination Survey (NHANES) is a program of studies designed to assess the health and nutritional status of adults and children in the United States. The survey is unique in that it combines interviews and physical examinations. The data files for more recent surveys are given in SAS Export format. To read these files in to R, one needs to use functions in the foreign package. If you don’t have this package, you may need to install it first. In the first step, we download these files and then in the second step we import these files to R.

# load foreign package (Converts data files into R)
require(foreign)    

# Set your working directory
setwd( "<YOUR WORKING DIRECTORY>")

### Download demographics file of NHANES 2005-2006 dataset
download.file(
ftp://ftp.cdc.gov/pub/Health_Statistics/nchs/nhanes/2005-2006/DEMO_D.XPT,
"Demo0506.xpt", mode='wb')

###Read downloaded file
Demo56<-read.xport("Demo0506.xpt")

### Download Blood pressure file of NHANES 2005-2006 dataset
download.file(
ftp://ftp.cdc.gov/pub/Health_Statistics/nchs/nhanes/2005-2006/BPX_D.XPT,
"BP0506.xpt", mode='wb')

### Read downloaded file
BP56<-read.xport("BP0506.xpt")

### Merge the two files
N_05_06 <- merge(Demo56, BP56, all=T)

You can download several files and then merge them together to get your dataset.

Updating R – in Windows 7

R is a great statistical software with tremendous flexibility. However, there is not a very straightforward (point and clinic) way of updating it. R Users have developed several different methods of updating R with its packages, including one described on CRAN.

I came across this one post, it is about updating R on Mac; tried it on Windows 7 with minor changes and it worked fine.

So here is what I did:
First, in the older version I wrote following commands
tmp <- installed.packages()
installedpkgs <- as.vector(tmp[is.na(tmp[,"Priority"]), 1])
save(installedpkgs, file="installed_old.rda")

Then I downloaded and installed newer version of the R. In the newer version of R I wrote following commands:
source(http://bioconductor.org/biocLite.R)
biocLite()
load("installed_old.rda")
tmp <- installed.packages()
installedpkgs.new <- as.vector(tmp[is.na(tmp[,"Priority"]), 1])
missing <- setdiff(installedpkgs, installedpkgs.new)
for (i in 1:length(missing)) biocLite(missing[i])

All packages were automatically installed to the newer version. Then, I went to Windows Control Panel and uninstalled the older version of R.

That’s it!

Some R Code Using SAScii

For example, I want use Adult Demographic File from NHANES III. To import it using SAScii

library(SAScii)
SAScode <- "ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/nhanes/nhanes3/1A/adult.sas"
ftpdata <-"ftp://ftp.cdc.gov/pub/Health_Statistics/NCHS/nhanes/nhanes3/1A/adult.dat"
data <- read.SAScii(ftpdata, SAScode, beginline=5)

Because the line with INPUT in the SAS code file begins at line 5, I gave the option begineline=5. Of course, it assumes that you have downloaded and installed the package ‘SAScii’. Now you can save this file in any desirable format or use it for further downstream analysis. It does take quite sometime, longer than what it would take using SAS, but it does produce desired output.

An Interesting R Package–SAScii

For several surveys, such as NHANES III, data for many files is available in ASCII format with SAS code. However, those of us, who want to use that data in R, it is little cumbersome to use first SAS, import data into it, then transfer data into a format that can be imported into R. I came across of this relatively new package, SAScii, that uses the ASCII format and then SAS code to directly import data into R. Quite nice.

REDUCE Trial – My Thoughts

This week, REDUCE trial was published in JAMA “Short-term vs Conventional Glucocorticoid Therapy in Acute Exacerbations of Chronic Obstructive Pulmonary Disease; The REDUCE Randomized Clinical Trial” JAMA. 2013;():1-9. doi:10.1001/jama.2013.5023.

REDUCE (Reduction in the Use of Corticosteroids in Exacerbated COPD), was a randomized, non-inferiority multicenter trial in 5 Swiss teaching hospitals that enrolled 314 patients (between March 2006 through February 2011) who had presented to the emergency department with acute COPD exacerbation and were past or present smokers (≥20 pack-years) without a history of asthma. Participants were treated with 40 mg of prednisone daily for either 5 or 14 days in a placebo-controlled, double-blind fashion. The predefined non-inferiority criterion was an absolute increase in exacerbations of at most 15% or a 6 months follow-up. The trial found that 5-day treatment with systemic glucocorticoids was non-inferior to 14-day treatment with regard to re-exacerbation within 6 months of follow-up but significantly reduced glucocorticoid exposure.

The trial results are interesting for those of us who actually practice medicine and see COPD patients on a regular basis with exacerbations. There are three settings in which the results of this trial can potentially impact practice. In office practice, there are definitely some patients who would do just fine with 5 days course of prednisone and these are patients who get prescribed Medrol dose pack. On the other hand, there are patients who need longer courses of steroids and for such patients it is important to give them longer courses of steroids (up to 14 days and sometimes even longer) to keep them out of hospital. Interestingly, based on only the severity of the patient’s symptoms and signs alone, it is impossible to predict who will need longer treatment. It is only history of the patients that tells what will work. The practice in Emergency Department is unlikely to be much different than in an office setting except that there may be patients with more severe exacerbation. There too, history is the only helpful thing. However, once patients are admitted to the hospital, it is likely that these are those subsets of patients who didn’t respond as quickly to steroids in ED and thus needed admission with persistent severe symptoms. For such patients, it remains a possibility that a larger number (if not all) of them will need longer therapy.

Thus, in my view, if a patient is new to me and presents with COPD exacerbation and I don’t have historical information on this patient, I will feel comfortable in giving this patient a 5-day course of steroids. Otherwise, if I have some additional information telling me that shorter course will not be helpful, I should go for longer course.

VPREB3 and Platelets

VPREB3 protein is the human homolog of the mouse VpreB3 (8HS20) protein, and is specifically expressed in cell lines representative of all stages of B-cell differentiation. It is also related to VPREB1 and other members of the immunoglobulin supergene family. This protein associates with membrane mu heavy chains early in the course of pre-B cell receptor biosynthesis. The precise function of the protein is not known, but it may contribute to mu chain transport in pre-B cells.

This protein doesn’t appear to be detectable in platelet proteome studies but its transcript is present in platelets (detected by both RNA-seq and microarray studies). Its role in platelet biology remains unclear. Even interesting is that the RNA-seq experiment found a much lower expression level than the microarray experiment (0.15 RPKM vs. 27250 MFI). Perhaps the level of gene expression is quite variable from person to person.

CD23 and Platelets

CD (Fc Epsilon Receptor II), has been shown to be present in platelets and may play a role in platelet aggregation. However, none of the publically available platelet proteome databases (Martens et al. Proteomics 2006; Burkhart et al, Blood 2012; Vaudel et al, Journal of Proteome Research 2012) have found this specific protein in platelets. When looking at transcriptome, CD23 RNA doesn’t appear to be present in megakaryocyte (using microarray). However, platelet RNA-seq analysis have found low levels of transcript in platelets (RPKM = 0.37).

While it is easy to speculate why there is such a discrepancy, it is possible that CD23 is induced in people with some allergy exposure and in individuals who are otherwise healthy (as were people in the studies that failed to find CD23), this transcript and its product may not be detectable.

It will be worthwhile to look at individuals with allergic responses (or parasitic infections) and examine whether they have higher expression of CD23 gene and protein. Comparing platelet aggregation in individuals with allergies and those without may also be illuminating.

Here comes STREAM ……

The results of STREAM were presented at ACC meeting and study was published online in NEJM – “Fibrinolysis or Primary PCI in ST-Segment Elevation Myocardial Infarction”. In nutshell the results can be summarized as pre-hospital fibrinolysis with bolus tenecteplase in conjunction with timely coronary angiography was similar to primary PCI in patients with early STEMI who could not undergo primary PCI within 1 hour after the first medical contact. Patients who failed fibrinolysis underwent emergent PCI (36.3% of the fibrinolysis group). Cardiogenic shock and congestive heart failure occurred more often in the primary PCI group and intracranial hemorrhage and ischemic strokes were more frequent in the fibrinolysis group. The study findings are likely to be reassuring for some parts of the world while may change treatment strategies at other places.

Some Useful Shell Commands for Cluster

bkill
especially bkill 0 as it will kill all jobs submitted by the user

bjobs

bhosts

Farnesyl Pyrophosphate and ADP-mediated Platelet Aggregation

Read this article recently and it seems interesting. Farnesyl pyrophosphate (FPP) can itself activate platelets and can induce platelet aggregation. However, this study concludes that FPP can act as endogenous antithrombotic factor by acting as insurmountable antagonist of ADP-mediated platelet aggregation. FPP is an intermediate in the cholesterol biosynthetic pathway. FPP also serves as a donor in post-translational isoprenylation of proteins. The steady-state plasma level was reported to be 6.6 ng/ml,  but  even  the  mild  physiological alteration caused by eating a meal has been demonstrated to increase the plasma concentration 200 fold. FPP is a natural antagonist of the LPA₂  (lysophosphatidic acid type 2) and LPA₃  receptors, and an agonist at the LPA₄  and LPA₅ receptors. While these receptors are present in platelets, FPP doesn’t appear to act through these receptors. Hogberg et al realized that the structure of ADP and FPP is similar as are their receptors. Thus, through a series of experiments they should that FPP inhibits platelet aggregation by blocking ADP receptors.

Platelet Function and Subsequent MACE in ACS patients

An interesting substudy of TRILOGY ACS was published in JAMA recently in which a group of patients with ACS underwent evaluation of platelet function after prasugrel or clopidogrel. All patients were also on aspirin. Thus, platelet function studies were performed after treatment with aspirin + prasugrel and aspirin + clopidgorel. TRILOGY ACS trial was a randomized, double-blind, active-comparator trial comparing prasugrel with clopidogrel in patients with unstable angina or non–ST-segment elevation myocardial infarction (UA/NSTEMI) who were managed medically without planned revascularization.Platelet function was assessed in a subset of enrolled patients using VerifyNow kits. VerifyNow P2Y₁₂ is a whole-blood, turbidimetric-based assay that measures platelet agglutination to fibrinogen-coated polystyrene beads after platelet activation with adenosine diphosphate. The results are expressed in PRU (P2Y₁₂ reaction units). There was no significant association between platelet reactivity and occurrence of ischemic outcomes during a 30 month follow-up period.

The results of this study are important; it is a large study, enrolling a high-risk population and followed for a relatively longer period of time. While trials of antiplatelet agents have established, beyond doubt, that platelet play an important (if not essential) role in the pathogenesis of ACS and perhaps also in the development and progression of atherosclerosis, studies have generally failed to find an association of platelet function with future events. Probably the most likely explanation is that we have, so far, been unable to identify a platelet function test that ACTUALLY measures platelet function in a way which is important clinically. We can predict bleeding but not aggregability. We do need to explore existing platelet function tests for association with future events that have not yet been examined and we also need to develop newer tests that are more closer in measuring what happens inside the vessel.

Today’s Interesting Links - September 10th, 2012

 

Genetic polymorphisms of CYP2C19 influences the response to clopidogrel in ischemic heart disease patients in the South Indian Tamilian population.

Regulated granule trafficking in platelets and neurons: A common molecular machinery

Idiopathic Sudden Sensorineural Hearing Loss: Classic Cardiovascular and New Genetic Risk Factors

Panobinostat (LBH589)-induced acetylation of tubulin impairs megakaryocyte maturation and platelet formation.

Itga2b regulation at the onset of definitive hematopoiesis and commitment to differentiation

Effects of clopidogrel added to aspirin in patients with recent lacunar stroke

WOEST - Optimal Antiplatelet Treatment in Patient Already taking Anticoagulants

Dual antiplatelet therapy (aspirin and clopidogrel or another P2Y12 inhibitor) is almost always prescribed (and is beneficial in preventing future adverse cardiovascular events) to patients with coronary artery disease who undergo coronary intervention and have stents placed (although it is associated with increased risk of bleeding). For most patients with atrial fibrillation or mechanical heart valves, anticoagulation therapy is a the standard of care (although it is associated with increased bleeding risk).

Question remains as to what to do with patients who are on anticoagulation but then develop coronary artery disease, undergo percutaneous coronary intervention (PCI) and stent placement. Use of dual antiplatelet therapy along with anticoagulation increases the overall risk of bleeding considerably and whether such a high risk of bleeding overshadows the benefit associated with dual antiplatelet therapy remains unknown.

In an ideal situation, one would randomize patients who are taking anticoagulants and who have undergone PCI with stent placement in one of the three arms: dual antiplatelet therapy, aspirin alone, or clopidogrel (or another P2Y12 agent) alone. Of course, all the three groups should continue taking their anticoagulation treatment. While, we don’t have such an ideal study, we do have results of a study (reported today in the ESC Congress 2012) in which patients taking anticoagulants were randomized to dual antiplatelet agents or clopidogrel alone. The results are interesting; not only that dual antiplatelet therapy in addition to anticoagulation was associated with increased risk of bleeding, it was also associated with higher risk of all-cause mortality (6.4% vs. 2.5%; p = 0.027).

One thing is certain from this study and that is that dual antiplatelet therapy, in addition to anticoagulation, is not optimal treatment and a single antiplatelet agent is likely to do a better job in reducing not only bleeding but also in decreasing all-cause mortality. However, whether this single antiplatelet agent has to be aspirin or clopidogrel, that remains unclear. In fact, it is quite possible that aspirin may have larger benefit than clopidogrel when used in such population. Perhaps future studies may be better able to point us relative benefits of antiplatelet agents in patients with PCI and stents who are also taking anticoagulants.

Clopidogrel vs. Prasugrel in Medically Managed ACS

Clopidogrel (famous by its brand name Plavix) is now generic and while prices vary from one pharmacy to another, it is very likely that very soon its price will come down. On the other hand, Prasugrel (which works the same way as clopidogrel) has a long way to go before it becomes generic and thus is likely to remain costly for next several years.

Prasugrel has one advantage over clopidogrel; its metabolism is not dependent on CYP2C19 (a member of cytochrome P450 system). A mutation within the gene of this particular enzyme decreases its function and decreases the conversion of the pro-drug, clopidogrel, into its active metabolite. The metabolism is shunted towards inactive metabolites and hence individuals with certain mutations in this gene are associated with increased platelet reactivity while on clopidogrel. This was shown conclusively in a GWAS by Alan Shuldiner and his team.

Prasugrel does seem to be superior to clopidogrel in individuals who undergo percutaneous coronary intervention (PCI) however, it was not clear if prasugrel is also superior to clopidogrel in individuals who are managed conservatively.

The results of the TRILOGY-ACS trial (double blind, randomized, double-dummy, active control, event-driven), announced today at the ESC Congress 2012 suggest that there is no difference. In this study, investigators randomized >9,000 patients with acute coronary syndromes (ACS) to either clopidogrel+acetyl salicylic acid (ASA or aspirin) or prasugrel+ASA. The primary end point of the trial was cardiovascular death, myocardial infarction, or stroke. The study did not find a benefit of prasugrel over clopidogrel for either the primary outcome or individual components of the primary outcome (p = 0.21). Post-hoc analysis suggests some interesting hypothesis that may be worth looking into in future studies.

Ticagrelor, another anti-platelet agent that also works by inhibiting the same ADP receptors (P2Y12), on the other hand, was shown to be superior to clopidogrel in a similar type of medically managed population in the PLATO trial. It will be interesting to see how they do in their head-to-head comparison (if ever done).

GWAS of Correlated Traits

Korte et al has written an interesting paper which extends the analysis of correlated traits to genome-wide association studies (GWAS). (Korte A, Vilhjálmsson BJ, Segura V, Platt A, Long Q,Nordborg M. A mixed-model approach for genome-wide association studies of correlated traits in structured populations. Nat Genet. 2012 Aug 19. doi: 10.1038/ng.2376. [Epub ahead of print] PubMed PMID: 22902788.)

GWA studies have commonly employed a simple statistical model in which a single locus is tested for association with a single phenotype (usually in an additive model). There have been few attempts so far, if any to utilize correlated phenotypes and perhaps improve power of the studies. 

In the 2011 meeting of American Society of Human Genetics at Montreal, Canada, I have used one, relatively simple, method to combine results from two correlated traits with resulting increase in power of GWAS. (Qayyum R et al. Correlated meta-analysis of genome-wide association studies of agonist-mediated native platelet aggregation in African Americans). What we did was to conduct two separate GWAS studies of correlated platelet aggregation phenotypes and then combined the results using meta-analysis. However, the resulting p-values were adjusted for correlation between the two phenotypes using tetrachoric correlation. We further conformed our findings using the simulations from correlated distributions, confirming our findings. 

Korte et al, use linear mixed model approach to not only adjust for population stratification but also for correlation between phenotypes. They use ASReml and R for this analysis and provide R scripts on their website. Using their techniques, Korte et al unveil additional SNPs in a GWAS of LDL and triglycerides.

 

Some Interesting Articles to Read

This one claims that odds are against men because of natural selection

Extreme phenotypes are helpful in identifying genetic loci and this study provides such an example

Here is introduction to Personal Genome Project, an interesting resource.

Ornithine Decarboxylase, Aspirin, and Platelets

Ornithine Decarboxylase (gene name ODC1) is the rate limiting enzymes of the polyamine biosynthesis pathway and catalyzes ornithine to putrescine. The activity level for the enzyme varies in response to growth-promoting stimuli and exhibits a high turnover rate in comparison to other mammalian proteins. Interestingly, this enzyme (as well as its inhibitor, Ornithine decarboxylase antizyme, gene name OAZ1) is also differentially expressed at higher levels in platelets from individuals with sickle cell disease than in in those without sickle cell disease (PMCID: PMC2225987). Platelets are in a basally activated state in patients with sickle cell disease and may contribute to at least some of the long-term vascular complications seen in patients with sickle cell disease. In patients with CAD, variants in OAZ1 have shown to be associated with increased risk of 6-month in-stent restenosis, increase in carotid intima-media thickness over the a 4-year period, and an increased risk of CAD (PMID: 17761941).

Aspirin is widely used for its anti-platelet effects and is also shown to be beneficial in reducing the recurrence of colon adenomatous polyps. A recent study has found that variants located downstream of the 3’ end of ODC1 gene may influence the risk of colorectal adenoma and impact the efficacy of aspirin. (PMID: 21930798)Whether there is a similar relationship between the antiplatelet effect of aspirin and ODC1 gene variants is unknown. Although one may be tempted to postulate that such a relationship exists based on the above noted studies, there are other potential mechanisms that may play a role. For example, ornithine decarboxylase has been shown to affect proliferation of vascular smooth muscle cells (PMID: 21894530) and the function of endothelial cells (PMID: 20594968); both cell types important in the process of atherosclerosis.

Selective Genotyping

In selective genotyping, a large number of individuals are phenotyped; however, only individuals with extreme traits are genotyped. This design tends to be more efficient than pure random sampling because phenotypically extreme individuals are likely to be genetically more informative [Huang and Lin, 2007].

Several regression-based methods have been developed for mapping QTLs under selective genotyping. These include the prospective linear regression [Xiong et al.,
2002], the retrospective likelihood approach [Wallace et al., 2006] and the conditional likelihood method [Huang and Lin, 2007]. Tang Y [2010] showed that the prospective, retrospective and conditional likelihoods actually yield identical score tests for association between a quantitative trait and a candidate locus.

Some Lectures on Basic Statistics

 

Statistical Reasoning I

Statistical Reasoning II

Introduction to Biostatistics

Above are three courses from Johns Hopkins Bloomberg School of Public Health