a question about annotation data in tophat

[bongbimit]

Let me start by saying I am very new to this so I may be asking naive questions

i am align my rna-seq reads (human) to genome by tophat but i am confusing about some parameter of annotation data in tophat

-G/--GTF : it mean i also have genes.gtf and i understand i know the exon, transcript and also the splicing junction

But
--no-novel-juncs: i dont understand what 's it function because i said above i have the genes.gtf file

thanks you all.

[dpryan]

A "novel junction" would be a new splice junction (i.e., a new connection between exons that doesn't exist in your GTF file). Finding novel junctions makes tophat take longer (it's slow enough as is, try STAR instead if you have enough memory) and may or may not be needed, depending on exactly what you're studying.

[bongbimit]

A "novel junction" would be a new splice junction (i.e., a new connection between exons that doesn't exist in your GTF file). Finding novel junctions makes tophat take longer (it's slow enough as is, try STAR instead if you have enough memory) and may or may not be needed, depending on exactly what you're studying.

but i think when i set parameter -G genes.gtf, tophat will map reads only into exons, other unmapping-reads will be slided smaller and map back into exon

so what is the different between from 2 parameter [-G --no-novel-junc] and only [-G]?

[blancha]

With the option --GTF or -G, TopHat will not only map the reads to the exons annotated in the GTF file. TopHat will first map the reads to the exons annotated in the GTF file, then TopHat will map the unmapped reads to the rest of the genome. You need to add the parameter -T/--transcriptome-only to map the reads only to the transcriptome defined in the GTF file.

The difference when you add the parameter --no-novel-juncs is that TopHat will not try to splice the reads at junctions other than those defined in the GTF file.

I generally set the parameter --no-novel-juncs since the runtime of TopHat is much longer without this parameter, as mentioned by dpryan, and there are more false positive mappings. However, if you are interested in discovering novel splicing junctions, not already discovered and annotated in the GTF file, you should leave this setting on.

[bongbimit]

With the option --GTF or -G, TopHat will not only map the reads to the exons annotated in the GTF file. TopHat will first map the reads to the exons annotated in the GTF file, then TopHat will map the unmapped reads to the rest of the genome. You need to add the parameter -T/--transcriptome-only to map the reads only to the transcriptome defined in the GTF file.

The difference when you add the parameter --no-novel-juncs is that TopHat will not try to splice the reads at junctions other than those defined in the GTF file.

I generally set the parameter --no-novel-juncs since the runtime of TopHat is much longer without this parameter, as mentioned by dpryan, and there are more false positive mappings. However, if you are interested in discovering novel splicing junctions, not already discovered and annotated in the GTF file, you should leave this setting on.

Thanks you all, i got it