Using a de-novo transcriptome assembled with Trinity, we are looking at differentially expressed genes between a control tissue and a tissue-of-interest in a spider.
Specifically, we are comparing the differentially expressed genes identified using Illumina paired RNASeq reads versus NuGen single 3' DGE System reads.
Currently, I am using Bowtie to map reads from each library type to the reference transcriptome and then using DESeq2 on the resulting raw counts to calculate differentially expressed genes.
Here is the experimental set up and the problem: we have 2 biological replicates for the tissue of interest for each of the read library types (Illumina and NuGen), but we only have biological replicates for the control tissue for NuGen reads. For the Illumina control reads, we only have one sample.
My problem is that because the Illumina analysis only has replicates for the tissue of interest and not the control, I'm not convinced I should use the statistical significance results from DESeq2. My limited understanding is that when there are experimental conditions where one has replicates and the other does not, DESeq2 calculates variance from the biological replicates that exist. However, it seems more reasonable to me to write the results based on abundance counts only and not the p-values.
So, two questions: 1. Should I ignore the p-values from DESeq2 because of my lack of control replicates in one of my analyses? and 2. Can I use the DESeq2 counts of log2fold change to compare the differentially expressed genes between by two library types?
Thanks very much for your time and help!
Specifically, we are comparing the differentially expressed genes identified using Illumina paired RNASeq reads versus NuGen single 3' DGE System reads.
Currently, I am using Bowtie to map reads from each library type to the reference transcriptome and then using DESeq2 on the resulting raw counts to calculate differentially expressed genes.
Here is the experimental set up and the problem: we have 2 biological replicates for the tissue of interest for each of the read library types (Illumina and NuGen), but we only have biological replicates for the control tissue for NuGen reads. For the Illumina control reads, we only have one sample.
My problem is that because the Illumina analysis only has replicates for the tissue of interest and not the control, I'm not convinced I should use the statistical significance results from DESeq2. My limited understanding is that when there are experimental conditions where one has replicates and the other does not, DESeq2 calculates variance from the biological replicates that exist. However, it seems more reasonable to me to write the results based on abundance counts only and not the p-values.
So, two questions: 1. Should I ignore the p-values from DESeq2 because of my lack of control replicates in one of my analyses? and 2. Can I use the DESeq2 counts of log2fold change to compare the differentially expressed genes between by two library types?
Thanks very much for your time and help!
Comment