De Novo Transcriptome Analysis To Get Differential Expression Data

Question

[nbvasani]

Hello User,

I need you with downstream analysis. I have fastq file containing sinlge end read from ion proton instrument. I have generated de novo assembly using Trinity. From there on wards, I am quite not sure how to get gene expression data?

Hope you guys can guide proper direction.

Thanks,
Naresh

[JC]

check http://trinityrnaseq.sourceforge.net/#Downstream_analyses

[nbvasani]

I read those Downstream_analysis, but it is not giving me good alignment percentage.

[JC]

in that case, your assembly is bad/fragmented, did you tried to do kmer-correction before assembly?

[nbvasani]

Nope. I just used trinity.pl script to generate de novo assembly. I don't think there is any option for kmer correction.

[JC]

there isn't in trinity, check for external programs. Related: k-mer correction in RNA-Seq data for transcriptome assembly?

[nbvasani]

Thanks a lot. I will definitely try that software.

Answer 1

[Charles Warden]

When using Oases for de novo assembly, I use eXpress for mRNA quantification. This should probably work with Trinity as well:

http://bio.math.berkeley.edu/eXpress/overview.html

After that, you can use your favorite differential expression tool (edgeR, DESeq, limma, etc.) on a table of normalized expression values from each sample.

[nbvasani]

Thanks cwarden45....

[nbvasani]

Hi cwarden,

I have question about creating reference de novo assembly using velvet/oases. I am planning to generate reference de novo assembly as follow: Hope I am following correct path. 1]Creating reference assembly by pooling reads of all 6-samples together in one file. 2] Run velveth with auto parameter for K from 27 to 71 on that sngle file 3] Afterword using velvethg for alleach k value 4] Running oases on contig.fa file generated from each K value 5] Transcript file generated from each assembly are merged using oases MergedAssembly script. 6] Checking merged assembly, by mapping each sample reads back to merged assembly using bowtie.

Am I on right path? Any suggestion would be appreciated.

Thanks, Naresh

[Charles Warden]

I would recommend using the Oases_pipeline.py script.

This allows you to run all of the steps in a single command. In fact, you can use it to run Oases / Velvet multiple times and combing the results (so you don't have to worry about finding the absolute best parameters to use).

For example, the following command will run Velvet / Oases with kmers set to 21 and 23:

python oases_pipeline.py -m 21 -M 23 data/reads.fa

http://www.ebi.ac.uk/~zerbino/oases/OasesManual.pdf