[Biopython] Lightweight version of Biopython?

Peter biopython at maubp.freeserve.co.uk
Wed Nov 3 08:38:29 EDT 2010


On Wed, Nov 3, 2010 at 12:16 PM, Erick Matsen <matsen at fhcrc.org> wrote:
> Hello Peter--
>
> Thanks for the quick response.
>
> Perhaps "heavy" isn't the right word, but it definitely seems overkill
> when all we want to do is convert between three file formats (FASTA,
> phylip, and Stockholm). Furthermore, something little we can include
> directly in the project (with credit given where it is due) would be
> ideal. An installation script could look for Biopython and use it if
> it's there, or use the lite version if not.
>

Hi Erick,

Under our licence you could take a snapshot copy of the bits you need
and include them in your application. However, the bits you'd need for
Bio.AlignIO would basically be the core sequence stuff in Biopython
(at very least Bio.Alphabet, Bio.Seq, Bio.SeqRecord, Bio.SeqFeature,
Bio.SeqIO, Bio.Align, and Bio.AlignIO). This adds up to a lot of files,
but would be possible... How are you distributing your application? If
using py2exe or a source code bundle you could in theory bundle in
(part of) Biopython.

If using pip or easy_install (i.e. PyPi packages) then it *should* take
care of a dependency on Biopython.

Suppose hypothetically with had a "lite" or "seq-core" version of
Biopython containing Bio.SeqIO, Bio.AlignIO and their dependencies
(but excluding Bio.PDB, Bio.Phylo, Bio.Cluster, Bio.Blast, BioSQL etc).
The download would be smaller (say a third of the size), but it isn't that
big to start with. There wouldn't be the (soft) dependency on NumPy.
Neither of these benefits seems compelling to me.

Regards,

Peter


More information about the Biopython mailing list