Finding Differentially Expressed Genes

Question

[DevS]

If I want to find the differentially expressed genes (DEGs) between two groups of cancers [say Cancer type I (it has sub types a, b, c) and Type II (it has sub types d, e, f) ], then what should be the approach and pls suggest what is the difference between the two approach I am writing below Approach 1- I take all the 6 cancers (a,b,c,d,e,f) and find the (DEGs) among them OR Approach 2- I find the DEGs for each cancer type (I and II) separately and find the common genes which are DE between the two groups

How are the two approaches different and which should be followed? Thank you.

[Sean Davis]

First, decide what biological question you want to ask. Between what groups do you want to find DEGs? After you answer that, it becomes easier to determine what tests to apply.

[DevS]

I want to differentiate between the two type of cancer based on genes which are common between them. In the first approaches i get genes which are DE across all the cancers (means these genes are common to all cancers). The second approach of finding common genes which are DE between two groups also gives me indirectly the DEGs which are common to all. I am little confused. pls help.

[Sean Davis]

Sorry, devyani. I still have no idea what you are trying to do. Perhaps someone else can help out.

Answer 1

[Sam]

Let me try to rephrase your questions: 1. Identify DEG that is common across different cancers 2. Identify DEG that's unique to the specific cancers

For the first questions, you might want to treat all cancer as case whereas the non-tumor samples as control, then perform the comparison. For the second questions, you will want to compare each cancer with their non-tumor samples to obtain all the DEGs, then once you've got the DEG list, you can see if any DEG is unique among all samples. You can actually also use this method to answer the first question, simply try and see if you've got any DEG that's identify in all samples. This have additional advantages of being able to identify DEG common in all cancer but with different directions.

[DevS]

Hey thank you Sam. I have already tried the second approach u have suggested to find the unique genes and and also find the common DEGs ( the first question). But the problem I faced was I didnt get any common DEGs (which I am interested in) among the different cancer samples. So I thought of little modifying my approach ( by not considering non tumor controls) and finding the DEs among only the cancer types. For this the two approaches I have mentioned in my question is what I could think of (of which I don't know which to consider). Later I will go for function analysis of the genes and consider only those which have role in cancer (as I have not taken control samples). I am new to microarray...so my assumptions and directions may sound illogical, but I am trying to understand. So any help will be very much appreciated! :)