I want to do a RNAseq analysis of Papilio.glaucus and now we have the sequences from 12 samples. I need a reference if I want to use Cufflinks/Tophat2. Should I merge my own transcriptome assemblies or use published genome data? How to evaluate a genome? I'm just starting and throwing out some basic questions I know...
The contig N50 of my own transcriptome assemblies are 969, 1009 and so forth.
Below are the info from published assemblies:
Papilio.polytes: scaffolds: 3,874 contigs: 14,375 N50: 47,768 L50: 1,129
Papilio.glaucus: scaffolds: 0 contigs: 92,145 N50: 12,225 L50: 8,335 (The species I'm working on)
Papilio.xuthus: scaffolds: 5,572 contigs: 10,777 N50: 128,246 L50: 528
Papilio.polyxenes densovirus: scaffolds: 1 contigs: 1 N50: 5,053 L50: 1
Thank you for any valuable suggestions!
Hi,
You can start by aligning your mRNA reads to each of the related species you mention using a more stringent criteria for alignment. This will let you get a feel for the quality of these genome assemblies and give you an estimate of how different your particular species is. You can also estimate the variability of this approach by checking the estimated expression values for each of the reference genomes used for a particular subset of shared genes. For example if expression values are always higher for one genome out of the four, you may want to leave it out of your analysis etc. Finally, the reads that do not align may represent novel transcripts and can be assembled de novo using one of the several do novo assemblers available, such as Trinity. Since you are only interested in evaluating the transcriptome, this approach may be sufficient. Best of luck!
Hi,
You can start by aligning your mRNA reads to each of the related species you mention using a more stringent criteria for alignment. This will let you get a feel for the quality of these genome assemblies and give you an estimate of how different your particular species is. You can also estimate the variability of this approach by checking the estimated expression values for each of the reference genomes used for a particular subset of shared genes. For example if expression values are always higher for one genome out of the four, you may want to leave it out of your analysis etc. Finally, the reads that do not align may represent novel transcripts and can be assembled de novo using one of the several do novo assemblers available, such as Trinity. Since you are only interested in evaluating the transcriptome, this approach may be sufficient. Best of luck!
Hi,
I would suggest you to generate the transcripts, for all of your 12 samples. This can be done by using velvet oases or using Trinity. In this process you will also get the information on different isoforms of the same gene. Once you generate the transcripts for all the 12 samples, merge them using cd-hit at 95% identity. This will allow you to generate the denovo reference for your genome of interest. Individual reads can be further aligned onto the reference, in order to study the expression profiles for different samples.
Best of luck!!
I would suggest to try both approaches as Suyash and Maxim suggest you. When I got the transcriptome from 16 samples, I played around using many references genome and tried to generate the transcripts of all my samples. On my case, the best assembly was the one using reference genome as the sequence genome was highly similar.
Good luck and let us know what works best for you.
All these suggestions are great. In addition, you could transfer the transcriptome from near species using RATT and combine it with your own assemblies with cuffcompare. The quality of your own assemblies could have great impact on the annotation. Cufflinks assembly may have problems with some organisms (I have worked with protozoa).
Dear Lu
May I suggest you reach consensus in using the alignment/statistical tools (Cufflinks, Bayesz, Isolator etc.). If you have any treated samples ( say pathogenic/mock etc.), please ensure you set the probability threshold to 0.05 and 0.01 just to find relevant transcripts that are across different treatments. This will allow you to do enrichment and annotation easily after hectic assembling.
Best
Prash
- Badges
- Science topic
- Similar topics
- Biological Science
- Evolutionary Biology