Tophat sorted bam error

[arcolombo698]

hello. thank you in advance.

I am receiving the error
open: No such file or directory
[bam_merge_core] fail to open file Sample_S6-N-2_output_single/tmp/accepted_hits0_sorted.bam


which causes an error and does not produce the accepted_hits.bam file which I need.


This was for a single read Tophat submission given for ONLY THE SECOND READ:

/auto/rcf-proj/sa1/software/tophat-2.0.10/bin/tophat --GTF /auto/rcf-proj/sa1/data/Homo_sapiens1/UCSC/hg19/Annotation/Archives/archive-2013-03-06-11-23-03/Genes/genes.gtf --num-threads 16 --library-type fr-firststrand --mate-inner-dist 100 --output-dir Sample_S6-N-2_output_single /auto/rcf-proj/sa1/data/Homo_sapiens1/UCSC/hg19/Sequence/Bowtie2Index/genome paired_trimmed_Sample_S6-N-2_R2.fastq.gz


any advice would be appreciated!

[yueluo]

Why use pair-end + stranded mode for a single read file of read2 ?

[arcolombo698]

Stranded Mode

Hello. THank you for the reply

the experiments that generated both reads used a stranded library, and thus I chose to use first strand because it is more accurate.

however, there were errors with the first read, so I had to separate the first read from second read.

but I still kept lib type - firststrand being that both reads were developed from stranded preperation type.

What would you advise to do? Do you suggest switching to lib-type unstranded? will this give better results for SE reads? Will this affect the data between library type ?

thank you again

[Brian Bushnell]

First strand is for single-ended libraries or paired libraries processed as pairs. Read 1 and read 2 come from opposite strands. So if you tell a program "first strand" and try to process read 2 as though it was single-ended, you'll get a wrong answer.

You should process the reads paired in all stages if you want to declare first strand or second strand; if you are going to separate them, then classify them as unstranded. Unstranded does not give worse results, it just loses the information of which strand the gene was on. Pairs, however, are MUCH more informative than single-ended for RNA-seq on organisms with splicing.

If there are errors in read 1, you can either use a more error-tolerant aligner, error-correct the reads, or trim, using a program that is paired-read aware to prevent discarding of one read in a pair. Coincidentally, I've written programs that can do all 3 things: bbmap.sh (error-tolerant rna-seq alignment), ecc.sh (error-correction), and bbtrim.sh (quality or adapter trimming), in the BBTools package.

[arcolombo698]

this is helpful and has answered my question. I was reading the Tophat manual and it really was not very clear about which strand corresponded to which read.
This site is helpful

http://onetipperday.blogspot.com/2012/08/three-ways-to-trim-adaptorprimer.html#uds-search-results