Factor out non-essential scripts and modules to removed dependencies

[AndreasHeger]

To reduce unnecessary dependencies, remove all scripts not in Manifest.in and their unique dependencies and separate them into a separate repository.

[IanSudbery]

For me the key dependencies to remove are:

• rpy2
• biopython
• corebio
• sklearn
• psycopg2
• MySQLdb

It would also be really nice if we could remove:

• pandas
• matplotlib
• bxpython

The following modules import the modules on the key list:

• Expression
• GWAS
• Biomart
• Counts
• CSV2DB
• Tree
• TreeTools
• RateEstimation
• Bioprospector
• SequenceProperties
• Stats
• Requirements (I guess this is a special case?)

Of these, only Stats and SequenceProperties are imported by the scripts. SequenceProperties uses the IUPAC codes, which I'm sure we can replicate.

The call to Stats.py in the scripts is to a single class Summary. While Stats.py does import rpy2, the Summary class doesn't use it.

[IanSudbery]

@AndreasHeger bxpython is an ongoing pain. For a start it is not going to be ported to py3.

In the production code it is used in:

• GeneModelAnalysis.pyx
• chain2stats
• gff2fasta
• gff2gff
• gtf2tsv

In those final three it appears that only bx.intervals.intersection is being used. Is there some way to avoid this? Is the algo used here much quicker than any of the others on offer?

[sebastian-luna-valero]

Since this repository has been refactored in https://github.com/cgat-developers/cgat-apps and bx-python as been replaced with quicksect, can we close this issue?

On the other hand, bx-python-0.8.1 has been migrated to Python 3 so we could add it back as it is not too dependency heavy (and available in conda)