cufflinks -- new transcripts detection

[capricy]

I have run tophat2 to align my paired-end rna-seq data to a reference genome:
/usr/bin/tophat2 -G ASgenome.gff3 -r 300 asgenome2 seq1.fa seq2.fa

and got results as ./tophat_out/accepted_hits.bam


Now I am reading the manual of cufflinks and have some questions about the option parameters:

1. what is the difference if I use -g ASgenome.gff3 or skip it? That is to say, will it be different between running

cufflinks -p8 -o clout ./tophat_out/accepted_hits.bam

and

cufflinks -p8 -g ASgenome.gff3 -o clout ./tophat_out/accepted_hits.bam


My purpose is to detect the new transcripts

2. cuffcompare results: class_code e

Should I remove those from final assembled geneset?

Any hints are welcome. Thanks a lot!

[mknut]

1. Yes they will be different. When you do not use -g the assembly is completely blind to the transcriptome information. When you supply transcriptome information with -g flag, it will be used to help guide the assembly. More details here.

2. It depends on what are you looking for. If you are focusing on novel isoforms you might want to keep them - they can originate from pre-mRNA (which is more likely), but they can also originate from mRNA showing different exon length than the reference. Keep in mind that this only applies if you used polyA selection in library preparation, while if you used ribosomal depletion, you should definitely ignore them (unless investigating pre-mRNA itself of course).

[capricy]

Thank you very much for your reply. I did redid cufflinks with -g.

Then I did cuffmerge and cuffcompare.


Further question about cuffcompare using the following command line:
cuffcompare -o ./merged_asm -i gtf_out_list.genome.2.txt -r ASgenome.gff3

gtf_out_list.genome.2.txt contains the information about the path to the merged .gtf


A bunch of files were generated in this step. I expected that .loci file gave the number of gene loci assembled, and this number should be the same as the number of genes derived from cuffmerge (genes.fpkm_tracking). However, the results showed that they are quite different. So which one should I go with? Could anyone explain why they were different?

Thanks for the help!

[capricy]

I am also confused about those .gtf files from cuffmerge (transcripts.gtf, merged.gtf) and cuffcompare(cuffcmp.combined.gtf) steps.

When I did cuffcompare, my input file (-i gtf_out_list.genome.2.txt) only has one file which is the transcripts.gtf from the previous cuffmerge. During the cuffmerge step, skipped.gtf is empty.

What I observe now is that they have different number of records.

Hope I can get some help here. Thanks a lot for your time!

[hubery_Bio]

When you run cufflinks with -g, it would conduct a RABT (reference annotation based transcript) assembly, and is helpful in discovering novel transcripts and isoforms.
But, I am confused about the cuffmerge and cuffcompare files, too. So, if you have any new opinion about these files, please tell me, thanks!