I am trying to align RNA-seq reads to a 'custom' mouse genome which contains an expressed human locus. To do this I concatenated the iGenome FASTA files for the mouse genome with the FASTA sequence of the transgenic locus. I also added in lines to the genes.gtf file matching the coordinates of the transgenes to the FASTA sequences. Chromosome names match between the two files. With the following input I get the error below:
tophat2 -r 30 --mate-std-dev 50 -G mm10_tg.gtf -p 15 --no-coverage-search --transcriptome-index=tg_transcriptome_data/known mm10_tg /wsu/home/ba/ba02/ba0287/TG_RNA/data/2M_test_whole_R1.fastq /wsu/home/ba/ba02/ba0287/TG_RNA/data/2M_test_whole_R2.fastq &
-----------------------------------------------
[2014-03-29 20:16:32] Checking for Bowtie
Bowtie version: 2.2.1.0
[2014-03-29 20:16:32] Checking for Samtools
Samtools version: 0.1.19.0
[2014-03-29 20:16:32] Checking for Bowtie index files (transcriptome)..
[2014-03-29 20:16:32] Checking for Bowtie index files (genome)..
[2014-03-29 20:16:32] Checking for reference FASTA file
[2014-03-29 20:16:32] Generating SAM header for mm10_tg
[2014-03-29 20:16:39] Reading known junctions from GTF file
[2014-03-29 20:16:43] Preparing reads
left reads: min. length=51, max. length=51, 1997863 kept reads (2137 discarded)
right reads: min. length=51, max. length=51, 1997632 kept reads (2368 discarded)
[2014-03-29 20:17:30] Using pre-built transcriptome data..
[2014-03-29 20:17:31] Mapping left_kept_reads to transcriptome known with Bowtie2
[2014-03-29 20:22:39] Mapping right_kept_reads to transcriptome known with Bowtie2
[2014-03-29 20:26:47] Resuming TopHat pipeline with unmapped reads
[2014-03-29 20:26:47] Mapping left_kept_reads.m2g_um to genome mm10_tg_ercc with Bowtie2
[2014-03-29 20:33:44] Mapping left_kept_reads.m2g_um_seg1 to genome mm10_tg_ercc with Bowtie2 (1/2)
[2014-03-29 20:35:58] Mapping left_kept_reads.m2g_um_seg2 to genome mm10_tg_ercc with Bowtie2 (2/2)
[2014-03-29 20:39:16] Mapping right_kept_reads.m2g_um to genome mm10_tg_ercc with Bowtie2
[2014-03-29 20:45:25] Mapping right_kept_reads.m2g_um_seg1 to genome mm10_tg_ercc with Bowtie2 (1/2)
[2014-03-29 20:47:39] Mapping right_kept_reads.m2g_um_seg2 to genome mm10_tg_ercc with Bowtie2 (2/2)
[2014-03-29 20:51:07] Searching for junctions via segment mapping
[FAILED]
Error: segment-based junction search failed with err =-11
Loading left segment hits...
Ideally I would like to make the above work if possible. But I have tried some other tests. The reads align fine to a w/t genome with supplied annotation (wt.fa and -G wt.gtf) and they will also align to the concatenated genome when no annotation is provided. I am worried with the latter approach that any conservation between the human and mouse genes will bias coverage towards the annotated mouse genes rather than counting as multiply mapped.
Has anyone ever attempted something similar? Thanks.
tophat2 -r 30 --mate-std-dev 50 -G mm10_tg.gtf -p 15 --no-coverage-search --transcriptome-index=tg_transcriptome_data/known mm10_tg /wsu/home/ba/ba02/ba0287/TG_RNA/data/2M_test_whole_R1.fastq /wsu/home/ba/ba02/ba0287/TG_RNA/data/2M_test_whole_R2.fastq &
-----------------------------------------------
[2014-03-29 20:16:32] Checking for Bowtie
Bowtie version: 2.2.1.0
[2014-03-29 20:16:32] Checking for Samtools
Samtools version: 0.1.19.0
[2014-03-29 20:16:32] Checking for Bowtie index files (transcriptome)..
[2014-03-29 20:16:32] Checking for Bowtie index files (genome)..
[2014-03-29 20:16:32] Checking for reference FASTA file
[2014-03-29 20:16:32] Generating SAM header for mm10_tg
[2014-03-29 20:16:39] Reading known junctions from GTF file
[2014-03-29 20:16:43] Preparing reads
left reads: min. length=51, max. length=51, 1997863 kept reads (2137 discarded)
right reads: min. length=51, max. length=51, 1997632 kept reads (2368 discarded)
[2014-03-29 20:17:30] Using pre-built transcriptome data..
[2014-03-29 20:17:31] Mapping left_kept_reads to transcriptome known with Bowtie2
[2014-03-29 20:22:39] Mapping right_kept_reads to transcriptome known with Bowtie2
[2014-03-29 20:26:47] Resuming TopHat pipeline with unmapped reads
[2014-03-29 20:26:47] Mapping left_kept_reads.m2g_um to genome mm10_tg_ercc with Bowtie2
[2014-03-29 20:33:44] Mapping left_kept_reads.m2g_um_seg1 to genome mm10_tg_ercc with Bowtie2 (1/2)
[2014-03-29 20:35:58] Mapping left_kept_reads.m2g_um_seg2 to genome mm10_tg_ercc with Bowtie2 (2/2)
[2014-03-29 20:39:16] Mapping right_kept_reads.m2g_um to genome mm10_tg_ercc with Bowtie2
[2014-03-29 20:45:25] Mapping right_kept_reads.m2g_um_seg1 to genome mm10_tg_ercc with Bowtie2 (1/2)
[2014-03-29 20:47:39] Mapping right_kept_reads.m2g_um_seg2 to genome mm10_tg_ercc with Bowtie2 (2/2)
[2014-03-29 20:51:07] Searching for junctions via segment mapping
[FAILED]
Error: segment-based junction search failed with err =-11
Loading left segment hits...
Ideally I would like to make the above work if possible. But I have tried some other tests. The reads align fine to a w/t genome with supplied annotation (wt.fa and -G wt.gtf) and they will also align to the concatenated genome when no annotation is provided. I am worried with the latter approach that any conservation between the human and mouse genes will bias coverage towards the annotated mouse genes rather than counting as multiply mapped.
Has anyone ever attempted something similar? Thanks.