Filtering with SignalP

• generate input file for online tool using "generate_SignalP_input_files" in the korbinian module: this will generate a file in FASTA format holding the accession numbers of the proteins and their full sequences (saved in "summaries/XX/ListXX_predictions/SignalP/ListXX_fasta_for_SignalP.txt", where XX is the list number (e.g. 02))
  • (see notes below if you have more than 2000 proteins in your dataset)
• navigate to http://www.cbs.dtu.dk/services/SignalP/ to access the online tool for SignalP

• Choose file : choose "D:\databases\summaries\XX\ListXX_predictions\SignalP\ListXX_fasta_for_SignalP.txt"

• change organism to match your input sequences

• change output format to (no_graphics) (short_no_graphics??)

• change cut-off values to User-Defined

  • 0.3 D-cutoff for SignalP-noTM networks
  • D-cutoff for SignalP-TM networks

• download the "gff.txt" output file

  • scroll down to the bottom of the SignalP output page
  • download "gff file of processed entries"
  • save in the summaries folder as "gff.txt", e.g. "D:\databases\summaries\XX\ListXX_predictions\SignalP\gff.txt"
  • korbinian will conduct signalp filtering if gff.txt file exists

• IF YOU HAVE MORE THAN 2000 PROTEINS IN YOUR LIST

  • Follow the instructions for Standalone SignalP setup
  • SignalP accepts a max of 2000 proteins in the web server
  • split your fasta file up into 2000-protein "chunks"
  • run signalP separately for each fasta file
  • join the output gff files into one big gff file (delete 3 header lines and paste each gff output underneath)