Consider samples from tissues A and B of a species X collected from geographical location 1 and samples from the same tissues A and B collected from geographical location 2. If we are to do differential expression analysis for all the samples, what will be the best way to generate reference transcriptome assembly? Pooling in reads from all samples together to do one transcriptome assembly OR performing assemblies for each sample separately and then pooling in the non-redundant assembled transcripts? Any other precautionary measures to be taken for such an analysis? Please share references if possible.
Hello Snehal,
I think it depends on what you want to do. Are you interested in identifying the differences between the tissues A and B between two geographical locations? Or are you interested to see the differences in each of the tissue individually when you harvested it from two geographical positions (i.e comparison of tissue A harvested from location 1 against tissue A from location 2; and the same for tissue B)? I think in any scenario, it would be better to do the assemblies individually (you can use Trinity assembler) and then move forward to the differential expression because like this you can make any comparison you want, but as I said it depends on what you want exactly to find out.
Hi Maria,
Thank you for your answer. We are interested in all sort of comparisons right now. I was wondering whether it is possible that assemblers like Trinity might provide wrong assemblies in cases where we combine reads and then do assembly. Also, instead if we do perform individual assemblies separately, and then combine into one master assembly, what sort of cut-offs should be used to remove redundancy (that will make sure that we are not combining two separate transcripts from two separate genes into one)? I am not exactly sure which is the ideal option here.
If you can share more information about why you would choose the later one and the preferable redundancy cut-off for it, then that will be really helpful.
Thank you and regards,
Snehal
Transcriptome analysis is always very specific type of analysis. Based on specific questions workflow can be changed. So all sort comparison using only one tool is not good option. Tissue/geography specific Assembly using different assembly algorithms and find best fit for your data, based on your questions will be most promising.
Hi Kiran,
Thank you for your views.
I could try different algorithms, but how would I be able to assess the quality of various assemblies in this case? I want to be sure that I am not making a mistake at the fundamental concept level before trying out multiple assemblers. Also even after choosing the assembler, the above question will still remain. How do I choose among the two - 1) pool reads OR 2) pool assembled transcripts, for generating reference/master transcriptome assembly (to which I can align reads and get FPKM counts)?
Populations from two geographical locations may possess allelic variations. How to handle those?
Regards,
Snehal
For the assembly, maybe you can have a look at this paper Article Combining Transcriptome Assemblies from Multiple De Novo Ass...
where the authors applied different assemblers for the de novo assembly and then they describe how they reduced the redundancy of assemblies and further steps until the annotation of the final transcriptome assembly. You can also check this paper Article Combining independent de novo assemblies optimizes the codin...
where the authors generated a concatenated assembly. A third suggestion is hereArticle De novo transcriptome assembly of RNA-Seq reads with differe...
where transcriptome assemblies were carried out with two RNA-Seq datasets generated from human brain and cell line and an efficient way to yield an optimal overall assembly was determined by using three different strategies.- Badges
- Science topic
- Similar topics
- Biological Science
- Anatomy
- Tissue