tophat taking too long to run

[fahim]

I am doing rna-seq analyses using tophat. The tophat is mapping against mm9 genome. It is almost 20 hours since I started the execution and it is still running.

Is it normal or I need to tweak any parameter to make it fast?

tophat -p 4 -G genes.gtf -o outputFiles/D13_178_MPP_CKO_tophatOut genome rnaseq/CKO/xxx_1_sequence.fastq rnaseq/CKO/xxx_2_sequence.fastq

[2013-03-19 15:14:14] Beginning TopHat run (v2.0.8)
-----------------------------------------------
[2013-03-19 15:14:14] Checking for Bowtie
		  Bowtie version:	 2.1.0.0
[2013-03-19 15:14:14] Checking for Samtools
		Samtools version:	 0.1.18.0
[2013-03-19 15:14:14] Checking for Bowtie index files
[2013-03-19 15:14:14] Checking for reference FASTA file
[2013-03-19 15:14:14] Generating SAM header for genome
	format:		 fastq
	quality scale:	 phred33 (default)
[2013-03-19 15:14:17] Reading known junctions from GTF file
[2013-03-19 15:14:19] Preparing reads
	 left reads: min. length=80, max. length=80, 40008050 kept reads (1239197 discarded)
	right reads: min. length=80, max. length=80, 40950279 kept reads (296968 discarded)
[2013-03-19 15:35:10] Creating transcriptome data files..
[2013-03-19 15:36:00] Building Bowtie index from genes.fa
[2013-03-19 15:39:59] Mapping left_kept_reads to transcriptome genes with Bowtie2 


[2013-03-19 17:41:34] Mapping right_kept_reads to transcriptome genes with Bowtie2 
[2013-03-20 03:43:06] Resuming TopHat pipeline with unmapped reads
[2013-03-20 03:43:06] Mapping left_kept_reads.m2g_um to genome genome with Bowtie2 
[2013-03-20 05:16:00] Mapping left_kept_reads.m2g_um_seg1 to genome genome with Bowtie2 (1/3)
[2013-03-20 05:20:00] Mapping left_kept_reads.m2g_um_seg2 to genome genome with Bowtie2 (2/3)
[2013-03-20 05:26:23] Mapping left_kept_reads.m2g_um_seg3 to genome genome with Bowtie2 (3/3)
[2013-03-20 05:42:15] Mapping right_kept_reads.m2g_um to genome genome with Bowtie2

[paolo.kunder]

is quite normal but it depends on your server specifications

[adiallo]

Try to use -p 8 is you have more power

Alpha

[Michael.Ante]

Build your transcriptome-index separately as mentioned in the manual:

tophat -G known_genes.gtf \
    --transcriptome-index=transcriptome_data/known \
    hg19

Afterwards you can use the transcriptome-index instead of the -G option. This will increase your computing time drastically.
You can use a subset of the reads to build it; according to the web-page, you don't even need reads to build the transcriptome-index with Tophat v2.0.10

[Brian Bushnell]

Alternatively, you can use STAR or BBMap, both of which are faster than Tophat.

[super0925]

Try to use -p 8 or -p 12 if you have enough power

[jernest1]

What computing system are you working on?

[blancha]

I always put --no-novel-juncs, unless the researcher is specifically interested in discovering novel junctions.
It's much faster, and results in less false positives that confuse the researcher