I try to call fusion junctions with MapSplice (1.15.2).

My input reads are 100nt in length and in fastq format.

Example:

@xxxyyyzzz
AGATGCACTACCACCCAGCCTATCAGGACAGGCCAAGCCTGAGGACAGTGACTGTCACAGAAAAATAGAAACTTGTGGTTCCAGGAAATCCGAGAGGTCT
+
IIHIHIHHIHHIHHGHHFGIGHGGGGIIIHHNKIJKIJELHCJFFGLIMLHLGLIAHBIEGFFDJQDAE>HIJHOIMLHAHGGHMPGHIFEJMDVJBEOM

I'm calling Mapsplice as follows (basically copy and pasted from the MapSpliceManual, just added the --fusion parameter):

python mapsplice_segments.py -u reads.fq -c [somewhere]/genome/hg19/singleChroms/ -B [somewhere]/genome/bowtie/hg19 -Q fq -X 5 -L 25 --fusion

MapSplice quits with an error:

[Thu Oct 25 11:24:45 2012] Checking for chromosomes files or directory
[Thu Oct 25 11:24:45 2012] Checking for chromosomes files or directory passed
[Thu Oct 25 11:24:45 2012] Checking for Bowtie index files
[Thu Oct 25 11:24:45 2012] check reads format
[Thu Oct 25 11:24:46 2012] merge paired end reads remove short
[Thu Oct 25 11:24:56 2012] Mapping reads against hg19 with Bowtie
[Thu Oct 25 11:25:13 2012] Converting bowtie mapped to SAM format
[Thu Oct 25 11:25:37 2012] divide reads
[Thu Oct 25 11:25:56 2012] Mapping reads against hg19 with Bowtie
[Thu Oct 25 11:26:50 2012] sort segmentbwt
[Thu Oct 25 11:27:15 2012] reads all chromo sizes
[Thu Oct 25 11:27:20 2012] mapsplice_search
[Thu Oct 25 11:30:06 2012] Aligning spliced reads
[Thu Oct 25 11:46:59 2012] mapsplice_report
error: open bwtmap file mapsplice_out/tmp/unspliced_mapped_segments.sorted error
        [FAILED]
Error: mapsplice_report failed

The file mapspliceout/tmp/unsplicedmapped_segments.sorted exists and is not empty.

When I turn of the --fusion parameter, MapSplice finishes with no errors.

Has anyone an idea, where the problem might be?