Entering edit mode
4.6 years ago
deniz
▴
10
Hi, I am trying to identify differentially expressed lncRNAs in tumor samples (18 replicates) compared to control samples (3 replicates). The same library prep kit is used for all samples; however, 10 samples have 100 bp paired-end reads and others 150 bp pe reads.
I am planning to use Cutadapt tool to trim the 150 bp reads to 100 bp. And after trimming, I am planning to use "Tophat>Cufflinks (with Gencode lncRNA annotation>Cuffmerge>Cuffdiff " pipeline.
I know that Cufflinks tool is deprecated. I’d like to improve my pipeline. Any suggestion will be appreciated.
Best,
Deniz
I don't see any issue with using both 100 and 150 bp reads together and thus no need to trim the 150bp ones back to 100 ?
how about the hisat-stringtie-ballgown pipeline ( the 'new Tuxedo' package)?
Thank you for the suggestions. Unfortunately, Ballgown is not implemented yet in the usegalaxy.org, and I am not experienced enough to the Ballgown analysis in R. The alternative analysis pipeline which also available in galaxy I am considering, is the "Hisat2-feature counts-DESeq2" pipeline. The tumor samples are FFPE(formalin-fixed paraffin-embedded) and normal samples are fresh. Do you think that the count normalization with DESeq2 will be efficient? Again, any suggestion will be appreciated.
never done this myself but that does sounds OK indeed.
just be sure that the lncRNAs are included in the gff files your provide to feature counts, otherwise you will not get any counts for those obviously .
more specific help you can also get from the galaxy help forum
Here you have a tutorial.. Tutorial HISAT SringTie Ballgown under R
In addition Nature´s detailed protocol