Tophat2 and hg19 - fail to generate fasta from indexes

E_w · 05-23-2014, 02:14 AM

Hi!

I just wanted to do a test run with Tophat2 but it appears more difficult than anticipated...
So I installed SAMtools and Bowtie2 in the required versions and then used the "make_hg19.sh" script from Bowtie to get the hg19 indexes. So far so good. But now when I run Tophat2 I get this:

Code:

tophat2 /usr/bin/bowtie2-2.2.2/indexes/hg19 FASTQ11_1.fq FASTQ12_2.fq 

[2014-05-23 09:56:35] Beginning TopHat run (v2.0.11)
-----------------------------------------------
[2014-05-23 09:56:35] Checking for Bowtie
		  Bowtie version:	 2.2.2.0
[2014-05-23 09:56:35] Checking for Samtools
		Samtools version:	 0.1.18.0
[2014-05-23 09:56:35] Checking for Bowtie index files (genome)..
[2014-05-23 09:56:35] Checking for reference FASTA file
	Warning: Could not find FASTA file /usr/bin/bowtie2-2.2.2/indexes/hg19.fa
[2014-05-23 09:56:35] Reconstituting reference FASTA file from Bowtie index
  Executing: /usr/bin/bowtie2-2.2.2/bowtie2-inspect /usr/bin/bowtie2-2.2.2/indexes/hg19 > ./tophat_out/tmp/hg19.fa
	[FAILED]
Error: bowtie-inspect returned an error
bowtie-inspect: reference.cpp:471: int BitPairReference::getStretch(uint32_t*, size_t, size_t, size_t, SStringExpandable<unsigned int, 1024, 2>&) const: Assertion `0' failed.

So it doesn't recognize the fasta files in that folder. Possibly because these are by chromosome and do not carry the same name as the index files. Eitherway Tophat2 should be able to reconstruct the fasta file from the index. Can anyone tell me where I went wrong?

GenoMax · 05-23-2014, 03:56 AM

Generally you would want to concetenate the chromosome files into a single "genome" fasta and then build indexes from that. Indexes consist of several files and it is conventional that they get the same "base name" as the genome file. (i.e. if your genome file was hg19.fa then the index files would have hg19.* prefix.)

If you are using standard Hg19 genome then there is no need to build the indexes/fasta file yourself. You can download an archive with these files (and more) along with the annotations from the iGenomes page: http://support.illumina.com/sequenci...e/igenome.ilmn

WhatsOEver · 05-23-2014, 03:59 AM

Is there a hg19.fa file in the folder where the bowtie index files have been created? Because this is required by bowtie. If you have it somewhere else, create a symlink in the folder to its location (from within the bowtie index directory: ln -s /path/to/hg19.fa ./hg19.fa). If you have the individual chromosomes as separate fasta file, combine in a single file using "cat ./* > ./hg19.fa" from within a folder where only your chromsome-fastas are in.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Tophat2: Ensembl GRCh37 V's UCSC hg19	SHeaph	Bioinformatics	2	01-10-2017 01:52 AM
tophat2 fail while running long_spanning_reads with out_of_range error.	julio.fernandez.banet	RNA Sequencing	1	08-09-2013 09:29 AM
Tools to generate VCF from two FASTA, or mutant FASTA from Ref FASTA and VCF?	jeffseq	Bioinformatics	3	05-28-2013 11:59 AM
Should I use chrM,random and haploid fasta file to build hg19?	mozart	RNA Sequencing	7	04-18-2013 01:07 AM
generate fasta file from chipseq peaks	says_anova	Bioinformatics	3	01-31-2013 09:40 AM

05-23-2014, 03:56 AM	#2
GenoMax Senior Member Location: East Coast USA Join Date: Feb 2008 Posts: 7,091	Generally you would want to concetenate the chromosome files into a single "genome" fasta and then build indexes from that. Indexes consist of several files and it is conventional that they get the same "base name" as the genome file. (i.e. if your genome file was hg19.fa then the index files would have hg19.* prefix.) If you are using standard Hg19 genome then there is no need to build the indexes/fasta file yourself. You can download an archive with these files (and more) along with the annotations from the iGenomes page: http://support.illumina.com/sequenci...e/igenome.ilmn

05-23-2014, 03:59 AM	#3
WhatsOEver Senior Member Location: Germany Join Date: Apr 2012 Posts: 215	Is there a hg19.fa file in the folder where the bowtie index files have been created? Because this is required by bowtie. If you have it somewhere else, create a symlink in the folder to its location (from within the bowtie index directory: ln -s /path/to/hg19.fa ./hg19.fa). If you have the individual chromosomes as separate fasta file, combine in a single file using "cat ./* > ./hg19.fa" from within a folder where only your chromsome-fastas are in.