Interpretation of differentially expressed genes results

Question

[M.Mohareb]

I have worked on rna seq and microareay public datasets. I have used Chipster Platform to reprocess the puplic datasets from fastq files or cel files to tsv files of differentially expressed genes(DEGs). The most important columns in those tsv files of DEGs of each dataset contain columns of p values, p adjusted values and log2fold change. I need to know how to interpret those data. I mean, what is the next step. How to find the most clinically relevant DEGs. Is there cutoff point to use the log2fold change. I mean, i have many DEGs with p values lower than 0.05, but many of those DEGs have fold changes less than one, minus or positive sign. I mean, -0.4, 0.66, or o.73 for example. How can i filter those data using log2fold change or p values. Thanks.

Answer 1

[JC]

How to find the most clinically relevant DEGs

You need to do pathway analysis or identify key genes for a particular disease, databases like OMIM has collections of known genes related to specific diseases. KEGG has collections of genes in disease pathways.

Is there cutoff point to use the log2fold change

In general, the adjusted p-value is more important than the FC, because it is derived from the p-value which indicates "what is the probability of this gene being differentially expressed?", but to compute good p-values you need multiple biological replicates. So, in many cases, we cannot trust so much the p-value, and a combination of an FC > |2X| is used in combination with the p-adj < 0.05. Some reports with large replicate numbers (>10) can just trust the p-adj and even FC > |1.1| can be considered trusted.

How can i filter those data using log2fold change or p values.

if you don't trust so much the p-values, use the FC > |2x| and p-adj < 0.05

[M.Mohareb]

Many thanks for your answer. But if i have adjusted p value <0.05 but log 2 fc <1 what will be the solution?? Please put into consideration that values are given as log2fc not fc. Should this be converted??

[rpolicastro]

A Log2 FC of 1 is the same as a FC of 2.