[tsucheta]

@Xinwu, My segment length was default 25.

[tonge]

@plabaj (and anyone else),
I have read your comment with interest and some concern. I plan to generate SOLID RNA-seq data (paired end 50bp+25bp) and wonder whether Tophat is any better at handling paired end SOLID data. I was planning to utilise Tophat for helping with new transcript/exon discovery. BTW this is all with mouse, so have annotated genome to work with.
Thanks.

[mackan]

Hi All,

I am trying to align paired-end Solid whole-transcriptome reads (50+35 FR) using Bowtie 0.12.7.

Strangely, when mapping paired-end, no read pairs (other than a few repeat-regions) map to the genome. In addition, no read pairs map to the transcriptome. (Ensembl genes-based reference).

When mapping individual reads, about 30% of 50bp reads and 20% of 35bp reads map successfully.

I must be doing something wrong. Using --ff changes nothing (reads are actually FR. The older Solid mate-pairs were FF) Read csfasta files are 100% pair matched (with read name suffixes _F3 and _F5-BC).

Examples (using only 1k reads but the results hold for the full set - as well as with exon-only Ensemble genes as reference, one fasta element per gene)
PE:
bowtie -f -p 8 --fr -C -S --sam-nohead --sam-nosq -s 1000000 -u 1000 --Q1
$dir/F3_QV.qual --Q2 $dir/F5_QV.qual -1 $dir/F3.csfasta -2
$dir/F5.csfasta ~/p2/indexes/bowtie/hg19_c align.sam
# reads processed: 1000
# reads with at least one reported alignment: 1 (0.10%)
# reads that failed to align: 999 (99.90%)
Reported 1 paired-end alignments to 1 output stream(s)

The aligned pair:
17_213_1598_F3 67 chr2 154876116 255 48M = 154876139 56 CACACACACACACACACACACACACACACACACACACACACACACACA LbRL\TMUVBH\TQ.8[[bOM```LG]^_LJ^\_SN]bcca_aaabbW XA:i:1 MD:Z:48 NM:i:0 CM:i:1
17_213_1598_F5-BC 131 chr2 154876140 255 33M = 154876115 -58 CACACACACACACACACACACACACACAACAGC @NcPIAE`c^E:S_^KI]`cPGZZSED)!E1!3 XA:i:0 MD:Z:29A3 NM:i:1 CM:i:5

F3-ends (50b) only
[markus@q34 run]$ bowtie -f -p 8 -C -S --sam-nohead --sam-nosq -s
1000000 -u 1000 -Q $dir/F3_QV.qual ~/p2/indexes/bowtie/hg19_c
$dir/F3.csfasta algn.sam
# reads processed: 1000
# reads with at least one reported alignment: 319 (31.90%)
# reads that failed to align: 681 (68.10%)
Reported 319 alignments to 1 output stream(s)

F5-ends (35b) only
[markus@q34 run]$ bowtie -f -p 8 -C -S --sam-nohead --sam-nosq -s
1000000 -u 1000 -Q $dir/F5_QV.qual ~/p2/indexes/bowtie/hg19_c
$dir/F5.csfasta algn.sam
# reads processed: 1000
# reads with at least one reported alignment: 287 (28.70%)
# reads that failed to align: 713 (71.30%)
Reported 287 alignments to 1 output stream(s)

Running tophat generated a decent result, so plenty of reads should map in pairs.

I'd be grateful if anyone can help me out, or provide any hint.

Edit: I got it working now, I might have forgotten about the --fr for the full data set.