Theodora, 

I would recommend that you look at iPathway-Guide (http://www.advaitabio.com/ipathwayguide.html#page=page-1).

The tools that most people mentioned above including GSEA, DAVID, etc. consider the pathways are simple sets of genes and completely ignore crucial information including the type of each gene, the position of each gene on each pathway, the type and directions of the interactions with the other genes, etc. In other words, the results of your pathway analysis are kind of independent of the pathways themselves. For most of these approaches, if you randomly permute the genes within a pathway, the results will be exactly the same. Furthermore, the approaches that calculate a p-value using a hypergeometric or Fisher exact test are very much flawed because these models assume the variables (genes) are independent. In fact, the pathways are there precisely to describe how these genes interact with each other in order to accomplish a certain function.

iPathway-Guide is the latest implementation of a new pathway analysis approach introduced a few years ago in Genome Research (http://vortex.cs.wayne.edu/papers/Genome_Research_reprint1537.pdf). This technique uses a systems biology approach that takes into consideration the direction and type of every edge on every pathway, the location of every gene, etc. This has been shown to both eliminate many of the false positives produced by the other approaches, as well as correctly identify true positives that are otherwise missed. This is why you may get very different results between different tools. There are 2 videos explaining this approach here (about 15 min each):

http://www.advaitabio.com/science.html

iPathway-Guide produces a “live” report that you can navigate and interrogate in various ways. This report includes a full GO analysis, the impact analysis as mentioned above, as well as the identification of miRNAs that are relevant in the given phenotype and the identification of putative mechanisms of action that would explain all measured gene expression changes. Links to drugs, diseases, literature, etc. are available for all entities (genes, pathways, etc.).

For a full explanation of the impact analysis, please see Chap 28 in:

http://www.amazon.com/Statistics-Microarrays-Bioconductor-Mathematical-Computational/dp/1439809755/ref=sr_1_1?ie=UTF8&qid=1422476470&sr=8-1&keywords=sorin++draghici

I would also recommend that you read Chap 24, Uses, misuses and abuses of GO profiling. A very common mistake (that many tools make) is to use the wrong background in both GO analysis and pathway analysis and calculate a p-value with a hypergeometric model. For instance, if you consider the background as being all genes in the human genome (about 47,000 on NCBI currently) while you only have 20,000 or so on your platform (Affy PrimeView), then all the p-values will be artificially more significant given a the warm fuzzy feeling, when in fact you might be looking at lots of false positives. The only way to do the analysis correctly is to consider the entire list of genes measured as the background and select from that the DE genes. Very few tools do that.

You can access both chapter for free on Amazon.

Sorin

Pathway Studio gives good results however this is not freeware.

Lots of good answers here from a previous thread.

https://www.researchgate.net/post/Can_anyone_suggest_Microarray_data_analysis_software_which_can_be_downloaded_free_of_cost_for_Windows_OS

Dear Theodora

I would certainly recommend Gene set enrichment analysis using GSEA

http://www.broadinstitute.org/gsea/index.jsp

and specifically performed Leading Edge analysis there afterwards tos elect the interesting genes

You can try also 1 month free trial of Ingeniuty pathway analysis (IPA). Its almost fully function but you can not download every report. It worked only for 2-3 times for downloading report (check ur inbox)

feel free to ask further queries

All the best

check this link for other DB and packages:

http://bioinformaticsonline.com/pages/view/17843/pathway-analysis

Have a look at the tools available here. Most of them are developed specifically for microarray analysis.

http://www.chibi.ubc.ca/faculty/paul-pavlidis/pavlidis-lab/software-and-resources/

WebGestalt, EnrichR and ConsensunPathDB are my favourites

I recommend MapMan, that is very useful if you need to display large datasets of expression data. Web page: mapman.gabipd.org/

There are a few around. If you are familiar with R and Bioconductor AND you want to take into account the topology of the pathways, I recommend the ROntoTools package. It implements the Impact Analysis described in the first publication attached, which is one of the few types of analysis that take into account the topology of the pathways. If you want something more elaborate, with great graphical capabilities, try iPathwayGuide at http://www.advaitabio.com .

A much older version of the software above is Pathway Express, found at http://vortex.cs.wayne.edu , search for "projects" and you will see the login link.

This last, however, is a commercial software, but I am sure that they will allow you a trial period during which you can test the full capabilities of the software.

A review of pathway analysis methods and software can be found in the second and third publications attached. 

Article A systems biology approach for pathway level analysis

Article Ten Years of Pathway Analysis: Current Approaches and Outsta...

Article Methods and approaches in the topology-based analysis of bio...

Hi, you can also use Babelomics (http://babelomics.bioinfo.cipf.es/). Check it out! ;)

Dear Theodora

I'm a fond user of Enrichr, which is free, easy to use and very efficient.

For some details, see a previous thread there : https://www.researchgate.net/post/Which_is_your_favorite_gene_list_enrichment_analysis_tool_and_why

In my hands, it worked best with lists of 150-800 (human) genes. With too few genes in your list, you loose sensitivity, with too many you lack specificity.

It's even more useful in connection with L2N, that allows you to save and manipulate your lists : http://amp.pharm.mssm.edu/l2n/upload/register.php

Good luck with your studies.

https://www.researchgate.net/post/Which_is_your_favorite_gene_list_enrichment_analysis_tool_and_why

http://amp.pharm.mssm.edu/Enrichr/

You may also try CytoScape. It is primarily built to be a network visualization tool, but there are plugins that you can download from within to construct networks. Both CytoScape and the plugins are free.

http://www.cytoscape.org/what_is_cytoscape.html

Dear Theodora,

if you don't mind to use a web-based tool I can highly recommend DAVID:

http://david.abcc.ncifcrf.gov/

You can use the the Affy IDs directly and check enriched Pathways (e.g. KEGG) or GO terms against all human genes or a pre-specified background.

Best regards,

Stephan

assume that you refer to 'pathway enrichment analysis'. several options are available:

1. gsea by sanger institute is a free jar software which can be used to do gene-set based enrichment

2. fisher exact test in R is another option in case you want to do it manually

3.  david functional analysis web interface

4. webgastalt, an alternative to david.

Please try - DAVID or WEBGESTALT

Hey...

I used IPA with trial version. I found that it's nice, easy, programming-free, not-so-complicated software/tool. But, it's trial version is highly restricted for use. In many places, you get message 'not possible with trial version'.

I would recommend use DAVID for Gene Ontology analysis. Others are GATHER, PANTHER. If you are looking for only pathway analysis, you can try STRING. All these are really simple. Pretty good to get first hand overview!

If you want to try a free "IPA-like" software tool I'd recommend STRING ( http://www.string-db.org ) to generate interaction networks but with a differential list of several thousand genes you will need to break this down to generate meaningful networks. You could try BioLayout Express3D ( http://www.biolayout.org ) for a free tool that uses Pearson correlation and Markov clustering to produce clusters of coexpression - this works very well on multi-thread/processor systems like a MacPro.

Try http://cbl-gorilla.cs.technion.ac.il/

The origianl paper is published in BMC Bioinformatics

http://www.biomedcentral.com/1471-2105/10/48

Good Luck!

I think you can go for cytoscope, string. If you want to use server you can try Kaas for smaller data set.

Hello Theodora,

We have exactly the same project ( Microarray data on Affy Primeview chip related to some cancer vs normal). As the chip is relatively new so you will find that many tools do not have readily available support for it like R does not have DB package for it,dCHIP does not work for its cdf file, David has issues in working with its probe IDs etc. etc which makes the analysis on this chip bit difficult and tricky to use these freely available tools.

I did build a pipeline with programming in R and perl and created our own DB and annotation code in R. Then mixed it quite a few tools like GSEA,EnrichR, BioPAX, Cytoscape, String and David refine and fine tune the outcomes.

As having the experience working on this chip, there is not a one simple way and tool to do the analysis efficiently. Need programming skills to tweak around this chip's data. If interested with our solution, let me know.

Surprisingly, no one mentioned TOPP suite, https://toppgene.cchmc.org/enrichment.jsp. Very simple, fast, and comprehensive, all results downloadable.

@Sanjeev: Good point one should always be critical to the results and understand what the enrichment analyses do. Here is the R function to perform simple GO and pathway enrichment using hypergeometric distribution, https://github.com/mdozmorov/R.genomerunner/blob/master/gene_enrichment.R

Theodora, 

I would recommend that you look at iPathway-Guide (http://www.advaitabio.com/ipathwayguide.html#page=page-1).

The tools that most people mentioned above including GSEA, DAVID, etc. consider the pathways are simple sets of genes and completely ignore crucial information including the type of each gene, the position of each gene on each pathway, the type and directions of the interactions with the other genes, etc. In other words, the results of your pathway analysis are kind of independent of the pathways themselves. For most of these approaches, if you randomly permute the genes within a pathway, the results will be exactly the same. Furthermore, the approaches that calculate a p-value using a hypergeometric or Fisher exact test are very much flawed because these models assume the variables (genes) are independent. In fact, the pathways are there precisely to describe how these genes interact with each other in order to accomplish a certain function.

iPathway-Guide is the latest implementation of a new pathway analysis approach introduced a few years ago in Genome Research (http://vortex.cs.wayne.edu/papers/Genome_Research_reprint1537.pdf). This technique uses a systems biology approach that takes into consideration the direction and type of every edge on every pathway, the location of every gene, etc. This has been shown to both eliminate many of the false positives produced by the other approaches, as well as correctly identify true positives that are otherwise missed. This is why you may get very different results between different tools. There are 2 videos explaining this approach here (about 15 min each):

http://www.advaitabio.com/science.html

iPathway-Guide produces a “live” report that you can navigate and interrogate in various ways. This report includes a full GO analysis, the impact analysis as mentioned above, as well as the identification of miRNAs that are relevant in the given phenotype and the identification of putative mechanisms of action that would explain all measured gene expression changes. Links to drugs, diseases, literature, etc. are available for all entities (genes, pathways, etc.).

For a full explanation of the impact analysis, please see Chap 28 in:

http://www.amazon.com/Statistics-Microarrays-Bioconductor-Mathematical-Computational/dp/1439809755/ref=sr_1_1?ie=UTF8&qid=1422476470&sr=8-1&keywords=sorin++draghici

I would also recommend that you read Chap 24, Uses, misuses and abuses of GO profiling. A very common mistake (that many tools make) is to use the wrong background in both GO analysis and pathway analysis and calculate a p-value with a hypergeometric model. For instance, if you consider the background as being all genes in the human genome (about 47,000 on NCBI currently) while you only have 20,000 or so on your platform (Affy PrimeView), then all the p-values will be artificially more significant given a the warm fuzzy feeling, when in fact you might be looking at lots of false positives. The only way to do the analysis correctly is to consider the entire list of genes measured as the background and select from that the DE genes. Very few tools do that.

You can access both chapter for free on Amazon.

Sorin

Problem is - just compare outcomes of these tools then will find very different outcomes. For example use David, GSEA , EnrichR and TOPP suite on same gene list and see.... Unfortunately many times people who have developed or are associated with some particular tool, recommend that. But without some programming and self intelligence- none of the tools works well to produce the problem specific accurate results. Customization of pipeline is must in biology as per the problem specifics.

I would like to thank everybody for your answers so far! 

I already tried iPathway-Guide and WebGestalt. As Sanjeev said, the results with the same gene list were pretty different! Unfortunately I have no programming skills, so R is not an option :(

@Sanjeev: I used the Affymetrix expression console for probe set summarization and CHP file generation and the Affymetrix Transcriptome Analysis Console for the identification of differentially expressed genes. Is there something wrong with this approach? 

DAVID is easy to use self-explanatory free online .

http://david.abcc.ncifcrf.gov/

GSEA is also good , but need more effort to use

http://www.broadinstitute.org/gsea/index.jsp

I would star with this two simple options. there are so many others. you can look them u in our paper: http://arxiv.org/abs/1411.0595

ingenuity used to be not very informative back in times. may be they are better now

Hello Theodora,

These Affy software (expression  & Transcriptome Analysis Console) uses RMA sort of analysis and which is mostly used with R also. So I think with proper threshold on Fold Change and p-value with your approach/tools  you should get diff. exp. gene list. Then you need pathway/enrichment/network/Go analysis whatever you need for you work.

Being in software/algorithms and looking at my biological problem, I write my code in R and perl and then combine with quite a few stats/tools/methods changing as per my data taking into account the empirical behaviors. At present I am working on this chip data for cancer cell lines. I would be happy to help/discuss with you and please feel free to write to me on sanjeevATbiocoslsDOTin

More focused on metabolism, you can use the web services from BioCyc (http://biocyc.org/PToolsWebsiteHowto.shtml#omicsDataAnalysis).

Several tools are available in this website for analysis of gene expression data, for example, the Cellular Overview enables you to browse through a metabolic map diagram that displays pathways according to groups based on biological function, or classes of metabolites produced/consumed.

By uploading a list of genes with their expression values, you can see which pathways/enzymes/transporters are up- or down-regulated and detect differences between the two conditions you want to compare.

gProfileR (http://biit.cs.ut.ee/gprofiler/) is very nice tool.

My favorite analyzing software is ingenuity IPA, however, it is not free...

candidates include but not restricted to: IPA, david, webgestalt, gsa, GSEA 

GeneSpring but its a commercial software

BRB Array Tools

-free, easy to use, has multiple tools, developed and supported by NIH employed bio-statisticians

MeV

-free, easy to use, has multiple tools, 

Pathway hunter tool.

Depends on your organism and how well annotated it may be. Some open-source all-around options would be STRING, STITCH, MANGO, and WGCNA.

If you don't have access to IPA, EnrichR is free and easy to use.

Here is a short list for pathway, but there are much more, I like David for giving the background,  EnrichR to give the option for saving and sharing the analyses and also for many annotation optioon,  ClueGo, BINGO in cytoscape for making better visualization. I work with IPA, but think it is really expensive

•GeneGo/MetaCore (www.genego.com)

•Ingenuity Pathway Analysis (www.ingenuity.com)

•Pathway Studio (www. ariadnegenomics.com)

•GenMAPP(www. genmapp.com)

•WikiPathways(www. wikipathways.org)

•cPath(cbio.mskcc.org/cpath)

•BioCyc(www.biocyc.org)

•Pubgene(www.pubgene.org)

•PANTHER (www. pantherdb.org)

•WebGestalt(bioinfo.vanderbilt.edu/webgestalt/)

•ToppGeneSuite(/toppgene.cchmc.org/)

•DAVID(david.abcc.ncifcrf.gov/)

•Pathway Painter(pathway.painter.gsa-online.de/)

EnrichR,

Hi,

Cytoscape software is free and you can use the ClueGo app to look for Go terms, Kegg and reactome pathways

As Sorin Draghici above pointed, all tools (that I know, including Reactome, DAVID etc.) except Ingenuity Pathway Analysis and iPathway, are actually enrichment tools, because directional or quantitative information between proteins is usually not available.

What is done in IPA and iPAthways but lacking in the open-source scene is actually a free directional (or even better quantitative) database, which would have to be curated by experts biologists.

Only with such a free open-access resource there could be development of open-source pathway analysis softwares...

Dear ALI,

Suggestion

Here is latest tools for expression analysis:

NetPath Signal

Pathway Commons

SPIKE

hiPathDB

IPA

PathVisio

Best,

ALI GHULAM

Dear Theodora

I prefer to use DAVID,

Good luck for your experiment

An updated list of suitable tools can be found here:

http://www.molbiotools.com/usefullinks.html#fungen

my choice is David

Clue GO plugin also nice Option in Cytoscape

Cytoscape

This article might help answers you question in a more practical way:

https://www.ncbi.nlm.nih.gov/m/pubmed/27899595/

Good luck

HZ

i have used this web based application for gene interaction network analysis.

you just have to put all gene ids.

http://www.enrichnet.org/

Website tool:

http://metascape.org/gp/index.html#/main/step1

Software:

Cytoscape and ClueGo

TopGo / edgeR/ limma - search in Bioconductor!

http://pantherdb.org/