Hi Everyone,
I'm using microarray data to identify DEGs and map its PPI network but now I want to use multiple datasets reported by different studies in Acute Myeloide Leukemia (AML). Please specify a good methodology step by step and also please specifically I can Merge different datasets. Please also need some info regarding the requirements for merging.
Thanking you in Advance.
This is extremely difficult, because you will have to consider differences in the protocols and also correct for batch effects. This is all a horror. If you are not really careful and "simply merge the data", you will get results that are either too noisy to be conclusive or that are more reflecting differences between the lab/protocols/batches than the biological differences you want to identify.
If you have no chance to avoid doing this, I strongly recommend working closely together with a bioinformatician and a statistician, who best are a bit experienced in such analyses.
Thank you very much Jochen Wilhelm but the thing is this is my research project and I need to go with this. if you could list some problems associated with such analysis I will try to solve those.
Collecting dataset having same GPL might work..but again the experiments conducted are under different experimental condition... so we have to very careful while doing analysis.... please follow the link below
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3348047/
https://www.researchgate.net/post/How_can_you_combine_different_published_expression_datasets_and_analyze_them_in_R
If the experimentation conditions are comparable, you can normalize each study separately.
Then use the MetaMA package of R to compute differential analyses on all studies.
The result is a list of DE genes taking into account the informations of all studies.
Becarefull, as said in the publication, it's recommended to perform meta analysis only on studies which use the same array technology. You shouldn't perform it on studies which use agilent array for one and affymetrix array for another for example.
Article metaMA : an R package implementing meta-analysis approaches ...
Hi agree with Jochen. This a very harsh task and highly error prone. Batch effects are magnified when comparing different experiments. Things are worst when experiments are performed in different labs.
I suggest to consider different analysis types than DEGs, and to study tools like ComBat.
https://www.bu.edu/jlab/wp-assets/ComBat/Abstract.html
Batch correction only works when the experimental conditions are (nearly) evely distributed over the batches. Otherwise, batch correction models will make things even worst (see link).
This is a severe problem when selecting data from different experiments, where many experimental issues are 100% confounded with the batches/experiments. If additionally biological groups are counfounded with experiments (like: I take group A from GSEx and group B from GSEy and compare A against B), the result is almost completely arbitrary (usually, results will already cluster nicely within experiments/batches, indicating considerable (but artificial!) differences in the expression profiles between the groups, but after batch-correction, these differences will be even exaggerated.
Further, if there are many different batches and relatively few samples per batch, a correction with combat will lead to an overestimate of the residual degrees of freedom for tests for DE (what might be corrected manually, but ignorance would be a bad guide). This prblem is also discussed in the attached link.
https://academic.oup.com/biostatistics/article/17/1/29/1744261/Methods-that-remove-batch-effects-while-retaining
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics