SpliceGrapher 0.2.0 Released

[CompBio]

We are pleased to announce the release of SpliceGrapher 0.2.0. With this release SpliceGrapher makes it easy to filter spliced alignments and to predict splice graphs:

• Simplified script for filtering spliced alignments in SAM files. SpliceGrapher now includes pre-built splice-site classifiers for over 100 organisms.
• Easy, self-contained script allows you to build your own classifiers.
• Now accepts gene models in either GTF or GFF3 format.
• Documentation now includes a Quick Start Guide and a User's Guide with a brief tutorial.

SpliceGrapher 0.2.0 is available for download from SourceForge

[bruce01]

Hi, I am trying to use Splicegrapher, especially the gene_model_to_splicegraph.py script, but I keep getting this error:

Traceback (most recent call last):
  File "/shared/SpliceGrapher-0.2.0/scripts/gene_model_to_splicegraph.py", line 23, in <module>
    from SpliceGrapher.shared.config     import *
ImportError: No module named SpliceGrapher.shared.config

Help MUCH appreciated, I'm sure it's something basic

[CompBio]

It looks to me like a path issue. You should be able to start python and type:

>>> import SpliceGrapher

If that doesn't work, then you probably need to add the location you specified with "setup.py install" into your Python path. For example, if you used the following install command

$ python setup.py install --home=/my/wonderful/stuff

...then you need to add /my/wonderful/stuff/lib/python to $PYTHONPATH.

I hope that helps. If not, feel free to e-mail me directly.

[Richard Barker]

hello
I'm also having issues with splicegrapher but at the next stage. So gene_model_to_splicegraph.py has split my custom GFF into chromosomes and the sam_collate.py command has split my sample.sam file into chromosomes. I'm now trying to use these files to generate splice predictions with the predict_splicegraph.py command but am having the following message.

What does it mean?

[user mydata]$ predict_splicegraph.py chrom_1_gene_model.gff -d 1.sam -M 12
00:23:01 predict_splicegraph.py Started
Traceback (most recent call last):
File "/usr/bin/predict_splicegraph.py", line 91, in <module>
origGraph = SpliceGraph.getFirstGraph(graphFile)
File "/usr/lib/python2.6/site-packages/SpliceGrapher/SpliceGraph.py", line 371, in getFirstGraph
result = SpliceGraphParser(f, **args).next()
File "/usr/lib/python2.6/site-packages/SpliceGrapher/SpliceGraph.py", line 1351, in __init__
self.loadFromFile()
File "/usr/lib/python2.6/site-packages/SpliceGrapher/SpliceGraph.py", line 1411, in loadFromFile
graph.addEdge(alias[e[0]], alias[e[1]])
KeyError: 'ENT1'
[user mydata]$

[CompBio]

The error arises from reading the GFF file, so that is the first place I would check, especially since you say it is a custom file. We have tested SpliceGrapher with a number of different GTF and GFF3 files, but yours may have a subtle difference that causes this problem. If you can describe how it was generated, we can try to address the issue.

[Richard Barker]

The original SAM file was generated with tophat2and bowtie2 using the following command.

[user]$ tophat -o output_name_bowtie2_for_splicegrapher --no-convert-bam -a 10 -g 1 -F 0 -p 6 --segment-length=28 --segment-mismatches=0 -i 40 -I 5000 --min-coverage-intron=40 --max-coverage-intron=5000 --min-segment-intron=40 --max-segment-intron=5000 TAIR10_chr_all.bt2.* -1 R1_Bowtie2_topat.fastq -2 R2_Bowtie2_tophat.fastq

The mydata.sam.gz was filtered with:
[user]$SAM_filter.py mydata.sam.gz alignment.sm classifiers.zip –o mydata_filtered.sam

The file was then split into 6 separate files based on chromosomes and the chloroplast:
[user]$ sam_collate.py *.sam
Generating: 1.sam , 2.sam , 3. sam , .4.sam, 5.sam , chloroplast.sam

I now want to use these files to describe the expression levels of the different genes/transcripts in cufflinks2 gene predicts &/or to generate new Splicegrapher gene models.
Cufflinks2 was then used to generate custom GTF files for 8 samples.
Cuffmerge was then used to generate a single merged.gtf files.

The merged.gtf file wouldn’t convert to a gff:
[user]$ gtf2gff3.py merged.gtf –o custom_merged.gff3

However, the original cufflinks gene.gtf could be converted to a gff3
[user] $ gtf2gff3 genes.gtf –o custom_genes.gff3

Or would it be better to use the transcripts file output by cufflinks?
[user] $ gtf2gff3 transcripts.gtf –o custom_transcripts.gff3

I used the custom_genes.gff3 to generate gene_models for each chromosome. Is there a command for the chloroplast and mitochondria?
[user] $ gene_model_to_splicegraph.py –m custom_genes.gff3 –c 1 –S –d /home/mydata
[user] $ gene_model_to_splicegraph.py –m custom_genes.gff3 –c 2 –S –d /home/mydata
[user] $ gene_model_to_splicegraph.py –m custom_genes.gff3 –c 3 –S –d /home/mydata
[user] $ gene_model_to_splicegraph.py –m custom_genes.gff3 –c 4 –S –d /home/mydata
[user] $ gene_model_to_splicegraph.py –m custom_genes.gff3 –c 5 –S –d /home/mydata
Creating: chrom_1_gene_model.gff3, chrom_2_gene_model.gff3, etc… 5 chromosomes in A.thaliana

I then tried to use the predict_splicegraph.py
[user mydata]$ predict_splicegraph.py chrom_1_gene_model.gff -d 1.sam -M 12
This was the command that gave the strange error message that is in my last post (see above).

Since then I have also tried to generate new splice variant pedictions using find_splice_forms.py command with cufflinks gene models to assist and got the following error:
[dr_richard_barker@vm142-21 mydata]$ find_splice_forms.py 1.sam -d /mydata/chrom_1 -m chrom_1_gene_model.gff
Traceback (most recent call last):
File "/usr/bin/find_splice_forms.py", line 90, in <module>
geneModel = loadGeneModels(opts.model, verbose=opts.verbose)
File "/usr/lib/python2.6/site-packages/SpliceGrapher/formats/loader.py", line 27, in loadGeneModels
return GeneModel(path, **args)
File "/usr/lib/python2.6/site-packages/SpliceGrapher/formats/GeneModel.py", line 754, in __init__
raise ValueError('No gene models found in %s' % gffPath)
ValueError: No gene models found in chrom_1_gene_model.gff
[user]$

I’m not sure what to try next?
Thanks for your help and advice...

[CompBio]

It appears that many of these problems are related to the way SpliceGrapher imports GTF and GFF3 files. We have recently made changes to these modules for version 0.2.1 that will be announced in a separate thread.

[Richard Barker]

The latest version is being installed on iPlant this week so i lookforward to using it...