Using Infernal to identify ncRNAs homologs

Install INFERNAL

• Go to http://infernal.janelia.org/
• Download The latest release of Infernal "infernal.tar.gz"
• Open a terminal window
• Unpack the source code:
  • tar zxvf infernal.tar.gz
• Change into the INFERNAL directory:
  • cd infernal-1.0.2
• Configure the installation:
  • ./configure --bindir=$HOME/bin
• Build the programs:
  • make
• Install the programs:

make install

• The programs we need are now installed in your "~/bin/" directory. There are some additional programmes from the easel package that are useful later in the exercise.

cd easel

./configure --bindir=$HOME/bin

make

make install

• Now change back to your working, download or Desktop directory:

cd ../../

Get an alignment that we can use

• Download an alignment from Rfam, for example PrfA thermoregulator UTR or choose your own by browsing Rfam.

Now we are ready to build a model!

Build a covariance model

• If you have RALEE and emacs installed, you can view your sequence alignment with pretty colour markup.
  • First rename your alignment something like:
    • mv RF00038_seed.txt RF00038_seed.stk
  • then:
    • emacs RF00038_seed.stk
  • Use the menus to colour the alignment by sequence conservation and then by structure.
  • Can you improve the alignment?
  • Quit from emacs.
• Otherwise, you might want to have a quick look at the alignment in your favourite text editor.
• Build a covariance model:
  • cmbuild -h
  • cmbuild RF00038_seed.cm RF00038_seed.stk
  • Take a look at the contents of the covariance model file if you want to:
  • less RF00038_seed.cm
  • (Hit "q" to quit from less)

Search a database of sequences

• Download a potentially interesting fasta sequence file from here. Save the file with the name database.fa or similar.
• Search the database using our model file:
  • cmsearch -h
  • cmsearch --tabfile RF00038_seed.tabfile -o RF00038_seed.cmsearch RF00038_seed.cm database.fa

While you're waiting, visit http://rfam.sanger.ac.uk/. Search or browse for the PrfA family. Have a look around at what is available.

cmsearch results

• Take a look at the results of the cmsearch:
  • less RF00038_seed.cmsearch
  • The raw data is perhaps not so informative if there are multiple hits to several sequences. You can view a tabular format:
  • less RF00038_seed.tabfile

Align the sequences with good scores

• Collect the sequences with scores greater than your threshold:
  • esl-sfetch --index database.fa
  • esl-sfetch --tabfile --Tmin 121 -Cf database.fa RF00038_seed.tabfile > new.fa
• Align the sequences to the model:
  • cmalign -h
  • cmalign -o new.stk RF00038_seed.cm new.fa
• Take a look at the alignment:
  • emacs new.stk
  • (First choose "Unblock alignment" in the "edit" menu to remove the alignment blocks)

Et voila!

We have built a covariance model, used it to identify putative PrfA elements and therefore a putative virulance gene, and aligned the PrfA element sequences back to the covariance model.