ESTpipe v0.4

ESTpipe is a pipeline for cleaning, annotating and submitting baboon
EST sequences into dbEST.

It is used in research for NIH grant "Baboon cDNA
sequencing" ( 1 R24 RR021863-01A1, CFDA No. 93.389,  2007-2010)

Many things have been hard coded in it in this initial submission, but
it can be easily modified to do process other ESTs.

The small test files that form part of the distribution, need to
written over by real data files. Once set up, the pipeline for one EST
library can be run by 'make'.

You can try:

  make -s; make clean

When make runs without error, you have installed all the dependencies
and can start processing real data.

ANALYSIS STEPS

1. Reject sequences with less than 64 informative residues after masking
2. Remove linkers from the ends (currently disabled because not needed)
3. Reject sequences with a match to human michondrial genome
4. Find a human protein homolog
5. Write out in dbEST submission format
6. Process the quality file to match rejected ESTs


DIRECTORY STRUCTURE

Since make is particular about identifying the files by their name and
location, it is important to understand the directories used:


data	    reference sequences
raw	    where EST files are placed for processing
work	    all intermediate files are created here
done	    final output files are place here

The flow of EST sequence data is:

  raw  -> work -> done

bin    programmes that make the pipeline
lib    library code for programmes
t      test for the library 
       (passing the tests depends on state of files in directories)



DEPENDENCIES

The following programs have to be installed for the pipeline to run

perl	      recent perl (5.8.0) is recommended
BioPerl	      use 1.6 release or a more recent SNV copy
	      http://www.bioperl.org/wiki/Getting_BioPerl
RepeatMasker  http://www.repeatmasker.org/RMDownload.html
CrossMatch    http://www.phrap.org
mdust	      http://compbio.dfci.harvard.edu/tgi/software/
NCBI BLAST    blastall and formatdb are used
     	      http://www.ncbi.nlm.nih.gov/BLAST/download.shtml

Note: If you have RepeatMasker/CrossMatch installed on other machine,
there is a trick for importing the processed sequence file.


PREPARE THE UNIPROT DATABASE

Copy the latest human swissprot and trembl files into data directory
using these names:

  data/uniprot_sprot_human.dat.gz
  data/uniprot_trembl_human.dat.gz

Make note of the release number for your records.

These files need to be preprocessed so that the header will contain
information about the source database.

To process these files and format the combined blast database run:

  make formatdb

The data directory also contains the human mitochondrial reference
sequence. If that needs to be updated, replace it with identically
named file (humanmito.fa) and this step will reformat that database,
too.


COPY THE EST FILES

The fasta and quality files for the EST library need to be placed in
the raw directory:

  raw/bab.fa
  raw/bab.qual

The quality file is not currently used by the pipeline, but is
processed to remove the sames sequences as from the fasta file that
were found out to be too short, repetitive or contain mitochondial
sequences.

The pipeline is dependent on finding meta information about the
sequence and the library from the fasta header. See the example fasta
file.


FIND REPEAT ONLY SEQUENCES

RepeatMasker/CrossMatch is run as part of the pipeline.

However, if you have it installed on a different machine, the
processed file can be copied into:

  raw/bab.fa.masked

This is the default setup. See the Makefile for details on
RepeatMasker command line options to use when running it.


RUN

At the root directory type: 

  make

Running removes all non-informative and contaminating sequences from
the final output.

The most time consuming step is the run the programme bin/estpipe that
does sequentially comparisons to reference sequences using BLAST using
BLAST. A detailed log is written to 'work/estpipe.log'. Any time
during the execution, you can analyse the log:

  work/estpipe_progress



OUTPUT


The output is written into the done directory. All files are named
according to the EST library name. This was parsed from the fasta header.

An example listing of the final files in the output directory, 'done':

  BABEVCEREB-C-01-1-7KB.dbest.1.gz  BABEVCEREB-C-01-1-7KB.dbest.7.gz
  BABEVCEREB-C-01-1-7KB.dbest.2.gz  BABEVCEREB-C-01-1-7KB.dbest.gz
  BABEVCEREB-C-01-1-7KB.dbest.3.gz  BABEVCEREB-C-01-1-7KB.fasta.gz
  BABEVCEREB-C-01-1-7KB.dbest.4.gz  BABEVCEREB-C-01-1-7KB.lib.gz
  BABEVCEREB-C-01-1-7KB.dbest.5.gz  BABEVCEREB-C-01-1-7KB.qual.gz
  BABEVCEREB-C-01-1-7KB.dbest.6.gz

The BABEVCEREB-C-01-1-7KB EST library has been written into
correspondingly named fasta, quality and dbEST files.

Importantly, the single dbEST file has been split into smaller files
each containing maximum of 10,000 entries that can be mailed as
attachments to dbEST ([email protected]).

Note that the dbEST lib file in this output was not created
automatically. You have to do it manually.  See the dbEST submission
documentation for details.


LICENSE

ESTpipe is licensed under the same terms as Perl itself, which means
it is dually-licensed under either the Artistic or GPL licenses.




Heikki Lehvaslaiho
20 April 2009