problem for bismark_methylation_extractor coverage output

[webappl]

I run bismark (v0.10.0) to analyse a single end directional methylation sequence dataset with the following commands:

(1) prepare genome file
bismark_genome_preparation --path_to_bowtie /opt/bowtie/ --verbose ./genomes/calbicans
(2) align with bismark
bismark -n 1 -l 50 --path_to_bowtie /opt/bowtie ./genomes/calbicans/ calbican1.fastq
(3) extract coverage report
bismark_methylation_extractor -s --bedGraph --counts calbican1.fastq_bismark.sam

The final calbican1.fastq_bismark.bismark.cov file reports very weird results. I list here the first 10 lines from the bismark.cov
Contig3 67 67 0 0 4
Contig3 153 153 0 0 2
Contig3 154 154 0 0 126
Contig3 171 171 0 0 2
Contig3 172 172 0 0 301
Contig3 280 280 0 0 1303
Contig3 308 308 0 0 2
Contig3 309 309 0 0 1380
Contig3 323 323 0 0 3
Contig3 324 324 0 0 1459

As you can see, unexpected outpus appear at positions 154, 172, 309 and 324. The large unmethylation count at position 280 is also suspicious. Looks like bismark messed up strand/direction.

What could be the problem occurring here?

Thanks.

[dpryan]

It's not clear from that that bismark did anything wrong in counting things. Have you looked at the aligned reads in IGV? Without seeing the raw alignments, we sort of have to assume that those numbers are correct.