Bioconductor - Science topic

You should be able to use cytofkit shinyapp without running any R code: just open R, and type library(cytofkit) and then cytofkitShinyAPP()

However, cytofkit runs on older versions of R (if I remember well, only R 3.6 or older versions).

Dear Dr. Ronán Michael Conroy , you are probably right. Thanks for the comment.

uninstall R and R studio

Then install R 4.1.2 https://cran.r-project.org/bin/windows/base/

Then install Rstudio 2021.0.9.1-372 https://www.rstudio.com/products/rstudio/download/#download

Have you tried to install it from the source file?

you're very welcome.

Retrieve that list of GEO accession and sample types

for ex:

Sample sample_type

GSM123444 Control

GSM123445 treatment1

GSM123456 treatment2

GSM123457 treatment3

GSM123458 treatment4

store the above list in a text file ex path.txt

Then import into R

data=read.table("path.txt", header=T)

Best!!

Pls go through our courses and let me know if you are interested to join

Bioinformatics :  https://www.eminentbio.com/application/views/Bioinformatics_course.pdf

Computational Biology: https://www.eminentbio.com/application/views/cb.pdf

R Programming: https://www.eminentbio.com/application/views/R_Eminent%20Bisciences.pdf

Python:https://www.eminentbio.com/application/views/Python.pdf

Statistics using R programming: https://www.eminentbio.com/application/views/Statistics_Rprogramming.pdf

Machine Learning using Python: https://www.eminentbio.com/application/views/ML.pdf

If you want join the course submit this online form : https://www.eminentbio.com/home/Onlineform

I met the same "FlowJo could not derive the expected parameter" problem.

After I assured R package flowAI worked in RStudio, I found out it's caused by my path of fcs files in Chinese.

So I changed related path in english, it worked!

Hope this will help you.

Have a good day:)

Sir can you share a tutorial or workbook or something like that. I can work in R, Python and Linux.

Thanks

Dear Riccardo;

A) the database of DAVID can convert the affy IDs.

B) If you are not skilled in programming:

1) download the latest annotation file for HG-U133_Plus_2 from the company website:

currently at: http://www.affymetrix.com/support/technical/byproduct.affx?product=hg-u133-plus)

find the file named

HG-U133_Plus_2 Annotations, CSV format, Release 36 (36 MB, 7/12/16)

2) unzip the file

3) open the file in MS excel

4) use "vlookup" function of excel to find the symbols. If you have a handful IDs you may search within the file.

If familiar with coding in R, then use annotationDbi (Bioconductor package)

Best

https://www.youtube.com/watch?v=lfzc_4oYow0

https://www.youtube.com/watch?v=EmZqlcKkJMM I hope this links will help to you Sunita Sharma

Thank you for your advice!

Nawaf Alomran Thank you for this information. This is a smarter way :)

Hi Ali,

first of all, I would suggest starting with a smaller dataset when learning how to analyse RNA-seq data. This will reduce a lot of headache and will give you an easier time bug-fixing and testing stuff.

Second, if you are serious about doing bioinformatics you will need to switch to Linux/MacOS at some point. You can keep Windows and install a virtual maschine running Linux to circumvent this. You will also need to dive deeper into bioinformatics to understand and build on what you are analysing (which is all possible but this goes beyond the scope of what you want to read here).

So to make it short, you can use galaxy and R for GE analysis of RNA-seq data. As you already have your data in galaxy, you can use "salmon quant" to quasi-map and quantify transcript levels (input are FQ files from one sample and a reference cdna you need to provide). Then you can use the salmon output to run DESeq2 and analyse differential gene expression between conditions (you need to specify you used salmon for mapping and provide a transcript ID to gene ID mapping file). Depending on what organism you are working with, it should be rather easy to find the cdna reference (just google) and a transcript ID to gene ID mapping file (you can use biomart, also just google).

Hope this helps!

Alexander

Dear Luiz Gustavo Nogueira Almeida the core of the limma package is linear model. A very simple explanation about why use model.matrix is that this function allow u to create several type of design matrix. According to the matrix created LmFit will estimates the parameters of interest.

So back to your example

data <- model.matrix(~factor(c(2,2,2,2,1,1,1,1)))

Means that you have two group, so when you fit your model you will estimates to parameters the first one will be the intercept that esitmates the population average in the first group and the second one that estimates the difference between the population averages of the second group and the first group. There is a whole chapter about design matrix on the book wrote by Rafael Irizarry and Michael Love called Data Analysis for the life Sciences.

THank you Ali!

The functional analyses show more or less the same pathways. Could this be indicating that the different bioinformatic approaches are not relevant for this case?

Please see

Hi Norman,

if you used the filters in GEO profiles and found nothing maybe you could go an other way. try in the pubmed query and see in found papers which one has data deposited in GEO...

fred

Thank you Frederic Lepretre and Maribel Baldellou for the advices..

My lab usually prepared two replicates for statistical analysis in microarray, and ran by the experienced technical staff.. but since this time the problem is about technicalities, we will prepare new samples and run the replicates at the same time.

As we need to have some hints from the DEG, for now we use FC>2 and will validate some using qPCR. I will design the primers around the probe.

Thanks again!

Yep! that´s make sense to me. Adjusted p-value does not have a big impact in such case but neither hurt. In my opinion, applying BH and selecting p-values above 0.05 as you did should be fine.

Best

Unfortunately, this is an issue with PC's when R packages (and dependencies) are downloaded to a user (not a root) directory... try "chmod -R 777" in the command line to allow read/write/execute permissions...

The 64-bit version of R should be able to adress all available memory, even under Windows (64 bit).

A possible issue is that you may keep several copies of the object in memory. Also, every time you make a change to an object, the entire object is copied. Make sure that you don't have any copies and call the garbage collector before starting a memory-demanding process.

However, if you anyway process it in a loop, you can divide it in chunks and process them one after the other.

And having a look at Emilys hint is also surely a good idea.

Thanks Sanjay. I ll let you know

Dear Glauco,

Hypermut, as suggested by Brian Foley, is to the best of my knowledge, quite good at estimating APOBEC-driven mutations, although it is more specific to A3G and A3F nucleotide contexts. The algorithm calculates the probability of a given seqeunce to be hypermutated by APOBECs by comparing the unbalance between mutated cytidines at APOBEC versus non-APOBEC (random) contexts, and provides a p-value related to that comparison.

Thank you for your answer and time Dr. Joachim. The link that you send helped me a lot and as you said the number 54675 probe set is true as well. Thanks again.

This should help https://laplacebayes.wordpress.com/2017/12/12/next-generation-sequencing-data-quality-checks/

Thanks mam. I have received data from DMET analysis. How will I present the data in various representations? What all software and applications will help in the same?

The current release of Bioconductor is version 3.5; it works with R version 3.4.0. Users of older R andBioconductor users must update their installation to take advantage of new features and to access packages that have been added to Bioconductor since the last release..

source: https://www.bioconductor.org/install/

regards,

Milan

I have some sample code here for a paper, where the data is downloaded from the GEO database and analysed using Limma. You can modify some sections, and use the Bioconductor Array Express package instead of Geoquery. 

Hope it helps. Good luck.

Thank you very much Mr. Kamstra and Dr. Muley! 

I am looking at human samples from infant cord blood lymphocytes. 

So far, I have a smaller list of CpG sites that are relevant based on p-values and fold-change (as what Dr.Muley suggested). 

Mr. Kamstra, I have associated CpG sites with genes using GenomeStudio, but would I have to use biomart to convert these gene names to ensemble IDs or entrez IDs? 

Thank you for your help!

I would like  to build on Farhad's answer. A good place to start would be the Cytoscape app store http://apps.cytoscape.org/apps/with_tag/ppinetwork where you find a list of  apps that deal with PPI analysis. You can see that the StringDB app is also listed here. However, I would recommend you to try StrongestPath and CytoGEDevo apps as well. However if you are more interested in doing a web-based analysis, you can always use PantherDB http://pantherdb.org/. It is simple, easy and intuitive. 

Good luck

Hi Mathavan,

If you are asking about this particular case, it can be explained by the fact that INO80B and WBP1 genes can sometimes occur as a read-through transcript that cannot be distinguished from either of the genes individually by the probes in question (see https://genome.ucsc.edu/cgi-bin/hgc?hgsid=579057461_Tm25CaPexEPPpKrMhRyrGSqaNmCR&c=chr2&l=74456725&r=74462493&o=74455022&t=74460891&g=refGene&i=NR_037849).

You can also look up more information about the U133 array's probes sensitivity and specificity to certain transcripts to understand why this could be for other probes and genes. For this particular example you can look at the information for probe 65133_i_at here https://genecards.weizmann.ac.il/cgi-bin/geneannot/GA_search.pl?keyword_type=probe_set_id&array=HG-U95&target=genecards&keyword=65133_i_at and for the 223072_s_at probe  see https://genecards.weizmann.ac.il/cgi-bin/geneannot/GA_search.pl?keyword_type=probe_set_id&array=HG-U133&target=genecards&keyword=223072_s_at.

I am getting "The page you were looking for was not found" error

Dear Denise,

the basic problem of high-througput data (common to metabolomics, transcriptomics and protein arrays) is the huge number of variables (protein species in your case) with respect to statistical units (the 14 samples) that open the way to a plethora of chance correlations. Thus the main lane IS TO EXCHANGE THE ROLE OF VARIABLES AND STATISTICAL UNITS. Simply operate on the transpose of tour original data matrix, i,e. the matrox having as rows the protein species and as columns (variables) the samples. On this matrix operate a Principal Component Analysis (that allow a dual representation of the same data set in terms of loadings (correlation coefficients of the variables with components) and scores (values of different component for each sample). So operate this pC, if you have 14 variables (7 control + 7 treatment) you will have in principle 14 components but, due to the mutual correlation between the abundance of different protein species you will end up into very few (2 or 3) principal acomponents explaining the by far major part of total variance.

Look at the component loadings and you will surely get a PC1 with all loadings of the same sign (size component) this is the signature of a common global profile shared by all the samples. Then go to PC2, PC3..PC4. You must see if there is a component in which the loadig values are significantly different between control and treated (ideally you will get  a component in which control and treated have opposite sign loadings) allowing for a perfect separation of the two groups in the loading space (shape components). If this is the case you will go to the scores of the discriminating component and look for protein species having the higher (in absolute value) scores on the component : those protein species are the ones allowing for the separation of the two groups and you solved your problem, If you will run a PCA on the correlation matrix you do not need to normalize protein values, normalization is implicit in the correlation metrics.

See:

Hi Ali, 

Thank you for your response. It is mRNA-Seq. I have edited the question.

Hello

Look at this link 

Good Luck 

Thank you Noha for your prompt response.  It was not quite what I was looking for (as I already know how to use some of testing tools) but I was looking for a package that could take data directly from the GSEmatrix and directly analyze that.  Thank you

Hi Neda

Since you refere to ArrayExpress I assume you will be analyzing microarrays - and for that the Limma package is the way to go - and it also have build in methods for doing gene-set and pathway enrichment analysis. A nice overview of the full workflow incl enrichment analysis can be found in this article http://nar.oxfordjournals.org/lookup/doi/10.1093/nar/gkv007 - and the Limma user guide is very nice and probably contain everything you need ( https://bioconductor.org/packages/release/bioc/vignettes/limma/inst/doc/usersguide.pdf ).

If you want to analyse RNA-seq you should go with DESeq or edgeR as previously suggested.

WARNING: You cannot directly compare different datasets from GEO or ArrayExpress (meaning comparing one cell type from one entry to another cell type in another entry). The problem is that the samples are created in different labs at different times (with different technologies) which by itself my affect the gene expression analysis. This problem is typically referred to as batch effects and means that you cannot conclude whether any differences you find are due to the cell types or due to the batch effect.

Hi Oveis,

If you plan to average the levels, I would use geometric mean to remove outlier influence.  However, since it seems like you are using the "s" set of affy probes (I did not see that when I first responded), they cross-hybridize to the same sequences as the design and it would be more difficult to determine biological impact as it is an inaccurate "group, probe, then analyze" method just using the R packages.

samtools mpileup pipeline or GATK might be useful. But don't sue them if you get any issue ;)

Thanks for your answers. All of them are valuable information for me. 

It's hard to say. A first guess wuld be that you should write edesign instead of design in the call to function p.vector. I'm not sure what the argument min.obs is for. But, maybe you have tor reduce the value to a number <= 18, since you have 18 arrays in your dataset.

hth

Matthias

I will use datatable if you have large matrix and you need speed. http://cran.r-project.org/web/packages/data.table/vignettes/datatable-intro.pdf .

exprmat_dt <- data.table(exprmat)

setkey(exprmat_dt,genename) # for sorting datatable

exprmat_dt[,list(mean=mean(expr),sd=sd(expr)),by=genename]

I assume that an absolute fold change is the one that uses ratios, rather than negative numbers. So an absolute fold change of 0.5 corresponds to a (conventional) fold change of -2. You take the negative reciprocal to convert from one to the other. However limma works with log 2 values which are negative when less than one. This is the usual way that limma does things. I'm afraid that you will have to try and figure things out from all this. But I'm am suspicious of the claim that they used limma for absolute fold changes.

My advice for anyone trying to trying to normalize data from different platforms is: DON'T! Although you can remove a lot of batch effects with ComBat, you are swamping the system with huge amount of technical variation: different samples, done in different places, done on different days, done on different technologies which is the worst type of technical variation of them all. The purpose of normalization is to remove technical variation and leave only biological variation, but you are throwing every conceivable piece of technical variation together. I personally would have no confidence in any result you derive from this and I suspect your combined gene list will be a piece of junk representing lists of genes could well correspond to the use of different technologies or different days of the week etc.

If you want to do comparisons between different experiments, analyse each of them individually and compare the resultant gene lists that you derive. What you lose in n-numbers you will more than compensate for with reduced technical variation. That's my advice anyway.

Have a look at the User's Guide for limma. It contains a description on how to use the function duplicateCorrelation() to address technical replicates.

Thanks for your nice reply, now you opened my brain to dig more in this topic, and as I told you earlier. I am totally satisfied by your answer it was totally more scientific than mine. I may expressed it in unspecified wording, and this may confuse the reader, but I meant the same answer you mentioned. But this gives me a nice lesson to not answer a question without writing it in a proper way (at least scientifically expressed, as you did) nice to know you and thanks for you, Raju, for bringing up this question.

There are several porcine annotation packages available:

Only the last three (porcine.db, porcinecdf, porcineprobe) are meant for Affymetrix arrays although not sure if they are compatible with yours. You may ask the Bioconductor annotation package maintainers at http://support.bioconductor.org

you have to consider this online course 

week 4 & 5

Have you tried this?

Your gene universe should really be limited to the original set from which your significant genes were taken.  For example, if your 28 significant genes were derived from an analysis of whole human genome microarrays, then you would limit your gene universe to just the ensembl gene IDs for the human genome (which is something like 20,300 or thereabouts).

Gene ontology hypergeometric enrichment is derived from basic set theory.  It is just the probability of k successes in n draws without replacement from a finite population of size N containing exactly K successes.  If you arbitrarily over-define one of the sets for comparison (K or N), you bias your enrichment results.

P.S. as an example of how to code TopGO to restrict your enrichment, here is an example from the Baliga lab at the Inst. for Systems Biology - http://baliga.systemsbiology.net/events/sysbio/sites/baliga.systemsbiology.net.events.sysbio/files/uploads/topGO_FunctionalEnrichment.r

The example looks for enrichment of a very small subset of human genes associated with glioblastoma (a type of brain tumors).

If you want to visualize the the change of GO enrichments through time you could consider Hive plots (using axes as time points and enrichment values on the axes).

Alternatively you can use D3 to achieve something like this: http://stackoverflow.com/questions/18569581/filtering-data-for-d3-js-sankey-diagrams

Hi Peter

"Thanks for the responses. I was hoping for a shortcut to get my analysis moving, but clearly I'll need to go do some more R homework."

There are some great free texts on the R-website that guide you through everything slowly (really great as reference material as well). Additionally Andy Field's book is amazing and will guide you through any basic analyses you need. Furthermore, try these resources for quick hints / help as in R you can get stuck on minor coding issues which they might resolve:

http://rprogramming.net/ - General help and topics

http://www.statmethods.net/ - great online guide and quick help options

Additionally, I would recommend downloading RStudio which will drastically increase its ease of use and power.

Thank you for your replies. I have studied few pappers about this issue starting from Michael Black suggestion. It turns that the cause of large amount of negative values is probably BeadStudio preprocessing (background subtraction) algorithm. Some publications inform that this procedure discards (by making values negative) many of significant intensity values. This preprocessing software has been used on those data and after that data was uploaded at GEO. It seems that the only way to work with this data is obtaining raw, non-preprocessed intensity values. Am I right?

You can find ready to use pd.mogene.2.0.st.v1 file at http://nmg-r.bioinformatics.nl/NuGO_R.html

Generally uses: TargetSCAN for miR gene annotations and then DAVID for functional clustering and annotations

Goodluck

I would assume that publishing on CRAN or BioConductor does not prevent you from publishing a journal article to help with visibility.

The Journal of Statistical Software (http://www.jstatsoft.org/) could be what you are looking for. The focus of the journal is not so much the method (which can be an established method) but the implementation. From this point of view you can release your package on CRAN and try to get more visibility through a paper in the journal.

Thank you Mr. Thomas Mohr. It worked !

Yes, basically correlate over/under expression of a given gene(s) to high/low CNV to specific chromosome band regions...

Once you obtain your DDCt values for each sample simply use the standard deviation function of Excel (separately for each treatement/genotype group). To get the SEM (standard error of the mean) divide the obtained standard deviation by the square root of the number of samples in the particular group. Remember that if you normalise the SEM has to be divided by the same value as your average was.This is how I was taught to do it, seems correct, but if it's not I'd be happy to hear how to do it properly.

I use GEDI and self-organizing maps because they are suitable to time-course and illustrative. The maps allow easily calculate entropy that describes organization of gene ensembles. I also like MeV.

Just a clarification:

Cufflinks can be used for building transcript models against which obtain counts with HTSeq-count. Our pipeline is: fastq+Tophat+ref genome>.bam files, .bam files+cufflinks> .gtf files (one per library)

.gtf files+reference transcriptome+cufmerge> merged.gtf file (merged annotation)

.bam +merged.gtf +HTSeq-count> annotated count matrix to input in edgeR or DESeq.

As it was said, the FPKM values from cufflinks can't be used in edgeR and DESeq because these packages analyze count data. In theory the FPKM values could be used with limma... after proper normalization. If needed would treat those values as expression levels from a single color microarray and apply some normalization. But I trully think that the nature of RNAseq calls for count-based models.

When if it still fails, check out the answers in this:

question 2.19 at http://cran.r-project.org/bin/windows/rw-FAQ.html

and answers here: http://stackoverflow.com/questions/6467277/proxy-setting-for-r

Bioinformatics and Computational Biology Solutions Using R and Bioconductor by Gentleman et al is a classic. It may be a bit long and outdated. I personally liked Bioconductor Case Studies by Hahne et al.

I could always take the help of really good books! :) Thanks a lot Laurin :)