A small collection of scripts to perform clustering of transcriptome-sequences.

• Predict open reading frames (ORFs) from transcriptome contigs using blast-hits as guidance
• Cluster orthologous ORFs of different species together
• Create peptide & nucleotide-alignments from those clusters

• (Bio)python for the pipeline itself
• OrthoMCL v2.0, which needs ** MySQL ** blastall-package
• Perl to run the protein2nucleotide-alignment-script

In short:

• Run Clustr.py and wait

In detail:

• Clustr.py will try to find a file named "config.cfg" to get all the settings, an example is config.cfg.example (surprise!). You can also specify the config-file using the "-c" parameter while starting: Clustr.py -c /some/config/somewhere.cfg
• In the config-file you will have to provide (looking at the example should give you an idea):
  • The organisms/assemblies you want to cluster
  • You can choose to skip the ORF prediction if you already have protein- & nucleotide-sequences for the ORFs
    • In this case: Specify the nucleotide-ORF-fasta as assembly_fasta and the protein-ORF-fasta as peptides_fasta
  • If you want to perform the ORF prediction: Where the blastx-supported protein-blast-database is located you want to use for the ORF-prediction
  • Where your makeblastdb-binary and the OrthoMCL-binaries are located
  • The credentials for your MySQL-database (write & delete-access is required)
  • Some output-details for OrthoMCL (name of the resulting clusters etc.)
  • After setting up your config: Run Clustr.py

• Implement further trimming of resulting alignments
• Implement analysis of resulting alignments

• Sometimes the nucleotide-sequences, given by the ORF-prediction, are longer than the corresponding protein-sequences. Due to this the protein2nucleotide-alignment won't work. Probably a bug in the orf-prediction-scripts.