splicing variant methods and their pros and cons
Hi All,
I read a post here about isoform expression quanification from RNA-Seq, and was really impressed by the summaries. I am working on a project comparing three methods (SplicingCompass, DEXSeq and MISO), for the exon-level differential exon usage. I have 8 biological samples in the control group, and 8 biological samples in the treatment group. The design is as below:
condition subject
C1 control 1
C2 control 2
C3 control 3
C4 control 4
C5 control 5
C6 control 6
C7 control 7
C8 control 8
T1 treatment 1
T2 treatment 2
T3 treatment 3
T4 treatment 4
T5 treatment 5
T6 treatment 6
T7 treatment 7
T8 treatment 8
As you can see from the last column, subject 1 has both the control and the treatment, and so on for all the other subjects. This is kind of different from conventional experiments where each sample receives its unique treatment. Our task is to detect differential exon usage between control and treatment.
Because of this and the fact that we have biological replicates, it seems that DEXSeq is the best fit in this situation (include subject as a covariate in the GLM setting). However, I just want to clarify
(1) if SplicingCompass can also handle biological replicates? It is less clear to me as users need to input GFF and BED files for each of the samples.
(2) if MISO is appropriate to use. This is a big concern because I see MISO has the biggest disadvantage of not handling bio. reps... It is really strange (as discussed in this post) that there are only two files: one control and one knockdown, to be passed to run_miso.py. What can we get (statistically and biologically) if we don't consider bio. reps?
(3) both SplicingCompass and MISO have to ignore the subject effect? I think there is no way the two methods can include the effect without using GLM... I really don't know how different the results would be.
Thank you so much for your suggestions!!
I read a post here about isoform expression quanification from RNA-Seq, and was really impressed by the summaries. I am working on a project comparing three methods (SplicingCompass, DEXSeq and MISO), for the exon-level differential exon usage. I have 8 biological samples in the control group, and 8 biological samples in the treatment group. The design is as below:
condition subject
C1 control 1
C2 control 2
C3 control 3
C4 control 4
C5 control 5
C6 control 6
C7 control 7
C8 control 8
T1 treatment 1
T2 treatment 2
T3 treatment 3
T4 treatment 4
T5 treatment 5
T6 treatment 6
T7 treatment 7
T8 treatment 8
As you can see from the last column, subject 1 has both the control and the treatment, and so on for all the other subjects. This is kind of different from conventional experiments where each sample receives its unique treatment. Our task is to detect differential exon usage between control and treatment.
Because of this and the fact that we have biological replicates, it seems that DEXSeq is the best fit in this situation (include subject as a covariate in the GLM setting). However, I just want to clarify
(1) if SplicingCompass can also handle biological replicates? It is less clear to me as users need to input GFF and BED files for each of the samples.
(2) if MISO is appropriate to use. This is a big concern because I see MISO has the biggest disadvantage of not handling bio. reps... It is really strange (as discussed in this post) that there are only two files: one control and one knockdown, to be passed to run_miso.py. What can we get (statistically and biologically) if we don't consider bio. reps?
(3) both SplicingCompass and MISO have to ignore the subject effect? I think there is no way the two methods can include the effect without using GLM... I really don't know how different the results would be.
Thank you so much for your suggestions!!