I have two conditions, untreated and drug treated. Some genes only express in response to drug treatment. So, I didn't get their expression in untreated condition and hence I could not calculate log2 ratio/fold change. But in pathway analysis, some programs ask for fold change value. How can I incorporate them into my pathway analysis along with other genes that have fold change value? I'm kind of new to microarray. Any suggestion would be highly appreciated.
What kind of microarray did you use ?
Don't you have background signal for unexpressed genes ?
Thanks David. I am using 3D-Gene DNA chip (TORAY). Yes, they have signal but I thought this might be the noise.
Ordinarily, you would use the normalized mean signal from your untreated samples, regardless of what that signal is. That represents your baseline signal for those genes, whatever it is (unless you feel you must exclude those as outliers, based on some rational assessment of outlier values).
You also do not need use any data values at all for enrichment analysis - you can base an ontology enrichment analysis merely on the lists of gene IDs determined to be differentially expressed. The rational being that you independently determine which genes are significantly differentially expressed and then simply look to see if that gene list is represented in a given ontology with a significantly greater than chance expectation.
yes, it's probably the noise, but this signal should be more or less the same for all unexpressed genes (or non-hybridized probes). I have no experience on 3D-Gene DNA chip, I work on Affymetrix. I think you can use this "noice" or "background" as a basal signal to calculate log2 ratio. That's what I do for Affymetrix microarray.
Thank you Michael and David. I have already done DAVID and Reactome analysis. Now I am starting Ingenuity pathway analysis which requires values. Thats why I was considering fold change.
Ingeniuty pathway analysis does not require any data values. I've used Ingenuity in the past (we now use GeneGO) and with either of those services, you can perform an enrichment with nothing more than a list of gene IDs. You can include data (fold change, normalized expression, p-value,...) but it is optional, not required.
Right - "Ingenuity doesn't require abundance vaues to calculate p-value for a pathway."
It never used to and it should not now. A hypergeometric analysis of two lists should require nothing but the lists themselves to compute p-values and adjusted p-values. The p-values and FDR are used to evaluate nothing more then the significance of the overlap in the query list with the defined ontology - no numerical data is required for such an analysis, just the lists of entities themselves.
Pathway enrichment, or any ontology enrichment analysis is nothing more complex then a hypergeometric comparison of lists. It is simply the probability of k successes in n draws without replacement from a finite population of size N containing exactly K successes.
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology