scanPAV

Pipeline to detect presence/absence variations (PAVs) when comparing two genome assemblies, assembly_1 and assembly_2.

To find PAV sequences present in in assembly_1 (presence-assembly) but absent in assembly_2 (absence-assembly), the pipeline performs the following steps:

       1. the presence assembly assembly_1 contigs/scaffolds are shred 
                  in chunks of 1000 bases after removal of Ns
       2. the 1000 bases chunks are mapped against the absence assembly assembly_2
       3. groups of consequent 1000 bases chunks that are not find 
                  in the absence assembly assembly_2 are merged and printed out as 'absence PAVs'

Download and Compile:

Requirements for compiling: gcc

	$ git clone https://github.com/SangerHpag/scanPAV.git
	$ cd scanPAV 
	$ ./install.sh

If everything compiled saccessfully you must see the final comment: "Congrats: installation successful!"

(Tested with gcc-4.9.2, gcc-4.9.4, gcc-4.8.1, gcc-6.0.2)

Test the installation using the yeast example:

	$ cd example
	$ ./run_yeast.sh

The test is using 20 cores and should take less than a minute. To modify the number of cores used by the test change the parameter "cpus" in run_yeast.sh before launching. The pipeline output will be in bwa_550/log.txt and the pavs in the same folder: bwa_550/. More details on the test in example/readme.txt.

External packages

The genome aligner BWA (http://bio-bwa.sourceforge.net) and SMALT (http://www.sanger.ac.uk/science/tools/smalt-0) are downloaded and compiled by scanPAV.

The package simulatesv (https://github.com/mlliou112/simulatesv) was used to create a modified yeast assembly to test scanPAV (in the example/ folder).

Run:

       $ /full/path/to/scanPAV/src/scanPAV -nodes <nodes> -align <aligner> -score <sw-score> \
   	      </full/path/to/assembly_1.fasta> </full/path/to/assembly_2.fasta> \ 
	      <pavs_present_in_assembly_1.fasta>
       
       where:
          /full/path/to/assembly_1.fasta: full path to the assembly file to be considered as "presence-assembly"
     	  /full/path/to/assembly_2.fasta:  full path to the assembly file to be considered as "absence-assembly"
     	  pavs_present_in_assembly_1.fasta:   output name for the pav sequences. 
     		These are the PAVs present in assembly_1.fasta, but absent in assembly_2.fasta
     
       parameters:
         nodes:    number of CPUs requested  [ default = 30 ]
         sw-score: smith-waterman alignment score [ default = 550 ]
         aligner:  sequence aligner: bwa or smalt [ default = bwa ]

Please notice that scanPAV technically only finds sequences present in assembly_1.fasta that are 'absent' in assembly_2.fasta. To find the sequences that are absent in assembly_1.fasta but present in assembly_2.fasta, scanPAV has to be re-run inverting assembly_1.fasta and assembly_2.fasta:

   $ /full/path/to/scanPAV/src/scanPAV -nodes <nodes> -align <aligner> -score <sw-score> \
   	      </full/path/to/assembly_2.fasta> </full/path/to/assembly_1.fasta> \ 
	      <pavs_absent_in_assembly_1.fasta> 
   
   	where: 	 
	   pavs_absent_in_assembly_1.fasta:  output name for the pav sequences. 
		These are the PAVs absent in assembly_1.fasta, but present in assembly_2.fasta

Results

The PAV sequences will be in the file pavs_present_in_assembly_1.fasta in your working directory. If you also run the scanPAV pipeline with assembly_1.fasta and assembly_2.fasta in reverse order, then you'll find also the file pavs_absent_in_assembly_1.fasta in your working directory.

Filtering of short PAVs

We recommend filtering out all PAV sequences shorter or equal to 1000 bp, as they are mostly noise and generally occurr on the edge of long variations or long N-gaps.

Some Notes on the aligners:

1. You can check if the bwa installation was succesfull by launching:

                $ /full/path/to/scanPAV/src/scanPAV-bin/bwa
   

   This should print out the bwa help information. If this gives you error messages, you can try to re-install it (by running again the install.sh script) or you can link your own bwa installation into /full/path/to/scanPAV/src/scanPAV-bin/bwa :

                $ ln -sf /full/path/to/my/own/bwa  /full/path/to/scanPAV/src/scanPAV-bin/bwa
   

2. The default aligner is bwa, but you also have the chance to use smalt, which is faster;

3. Results from smalt and bwa are relatively consistent, but some small differences are to be expected due to different aligner sensitivities and internal parameters.