[Pypedal-devel] Gianluca to John

[John C. <joh...@gm...>]

Alle 21:36, mercoled=EC 01 settembre 2004, John Cole, ha scritto:

> I have attached a file, pyp_newclasses_gianluca.py, with responses to
> your questions.  They are denoted "JB:".  Let me know if I failed to
> answer your questions clearly.  Please remember that I have only
> worked on pyp_newclasses.py a little bit -- the code does NOT work yet
> and there are lots of missing pieces.

I'm sorry if I appear a little boring, but in this way I can understand you=
r
programming point of view...

>  Also please bear in mind that
> the code evolved to meet my personal needs, so my design decisions may
> not always make sense.  There were a lot of deadlines that I had to
> meet to get my dissertation done in time to graduate.  :-)  I am going
> to create a developer forum on the Sourceforge site -- I think that it
> would be better to have these design discussions there so that I do
> not lose any ideas.

I'd prefer to use the mailing list. I've still subscribed it.

> As far as spaces versus tabs goes, I do not really care.  If you
> prefer spaces then we will use spaces.  You will see that there is a
> mix of the two styles in the CVS tree; I keep trying different editors
> and cannot find one that I really like.  Maybe I should try WingIDE.

Did you try eric3?
http://www.die-offenbachs.de/detlev/eric3.html

It's a little boring to install, 'cause it runs with qt and needs also
qscintilla, sip, pyqt, but when I tried it, I really SAW THE LIGHT!

       # JB: At the moment source is not checked until the file is opened (=
if
if exists), so for now it assumes that the
       #     user will know what to do when Python throws a file-not-found
exception.  We can check it here to save a little
       #     time, I guess.
       # JB: As far as source being multiple files...we would have to do a
little more work to make sure that there are no
       #     animalID collisions between the two files but I do not see any
reason why pedigree data could not be loaded
       #     from more than one source.  This would require multiple calls =
to
load() but that is doable.

When I talk about multiple sources I mean databases, plain text files, gedc=
om
files, xmls, pickled files and so on. Each one store pedigrees in a differe=
nt
way. For this reason is better to do a Pedigree class that doesn't care abo=
ut
pedigree format. if source is a list, you can define a factory function lik=
e
those:

def pedigree_from_file(inputfile, format, ecc.):
 list=3Dsome_operation_on_input_file()
 return Pedigree(list, other_arguments)

def pedigree_from_db(driver, user, passwd, db, sql_statement):
 list=3Dsome_operation_on_db()
 return Pedigree(list, other_arguments)

       # GS: Here all arguments specified in kw becomes class attributes:
(python 2.2 and later)
       # JB: I have not used the **kw construct before, but I want to.  If
you look at some of the
       #     procedures in pyp_utils.py you will see that a ridiculous numb=
er
of aprameters are
       #     passed explicitly.  Using **kw is a little bit better way to
handle that.

you only save some parameter declarations. but in taht way you make your co=
de
harder to understand

       # So let's learn how to read things in from a configuration file.
       #GS: for such kind of operations I prefer to use XML file. Xmls are
harder to modify by and, by easier to
       #GS: modify by a routine. Please, remeber me to send you a little
class to manage configuration xml.
       # JB: This is probably a matter of personal preference.  I feel that
an XML configuration file is needlessly
       #     complex.  In fact, I feel strongly enough about that to insist
on using ConfigParser OR to support both
       #     styles of configuration file.  You can try and change my mind =
if
you like, but you probably will not
       #     succeed.  :-)

I attach two files, an xml and a little class to parse it. To support both
options is not a good idea. decide what's the better for you; however this =
is
a secondary aspects.

           #GS: From where renumber comes? have you forgotten module
reference?
           # JB: This is NOT working code.  The idea is that renumber will =
be
moved from
           #     pyp_utils.py to a method of this class.

Ok, sorry. Regarding this point. I thinked that there are two different
strategies in structuring classes of pypedal. One is to build a pedigree
class that contains all calculation methods. The other is to build a pedigr=
ee
class that only store a pedigree and different classes that take a pedigree
instance as __init__ argument and perform specific calculation.

I make a little schema to better understand myself :-))

class Pedigree:
 def __init__(self, arguments...):
    methods and attributes

class demog:
 def __init__(self, PedInstance, ecc)
   ...
 def age_distribution(self):
   pass

class metrics:
 def __init__(self, pedinstance):
   ecc

__init__(self,animalID,sireID,damID,gen=3D'0',by=3D1900,sex=3D'u',fa=3D0.,n=
ame=3D'u',alleles=3D['',''],species=3D'u',breed=3D'u',age=3D-999,alive=3D-9=
99):
       #GS: Why to set by=3D1900? In humans pedigree this can provide some
matters. I'd prefer None
       # JB: This is an arbitrary number.  The code was originally written =
to
work with dog pedigrees that had complete
       #     birthdate information.  If we set it to None then we will need
to trap that in some other procedures/methods.

Than -1 can be a reasonable value? So animals without "by" in sorting pedig=
ree
comes allways before all others

       #     For example, the Boichard algorithms for effective founder and
ancestor numbers require generation information.
       #     When generations are not provided in the pedigree file they ar=
e
inferred using, in part, birth year.  I am open to
       #     changing this.

Have you an article or a something else that can be attached in an e-mail t=
hat
can me explain the Boichard algorithms? I googled for that but I didn't fin=
d
nothing.

       # JB: Oh!  The pad_id() method uses the birthdate to form the padded
ID.

A question, what is the pad_id() function intended for?

       #GS: As soon as I can understand only two alleles can be specified.
       # JB: Because I only added alleles to support gene dropping.  I am
open to changing this as well --
       #     human geneticists probably need to handle more information tha=
n
that.

At the moment there is not need for managing multiple markers information, =
but
in the (far?) future the class pedigree can be extended to perform some
calculation on genotype, for example haplotype extimation, linkage
disequilibrium etc...
so can be useful a structure like a dictionary:

genotype=3D{'markerName':["allele1", "allelee"],'anothermarkerName':["allel=
e1",
"allelee"],...}

       if alleles =3D=3D ['',''] and self.founder =3D=3D 'y':
           _allele_1 =3D '%s%s' % (self.paddedID,'__1')
           _allele_2 =3D '%s%s' % (self.paddedID,'__2')

here is an answer for the question above.

       #gs: This  concerns the feature I ask for. Do you think it is possib=
le
to use STRING animal's(person's) ID?
       #gs: perhaps with a mandatoru renumber.
       # JB: I think that it would not be too hard to do that.  We just nee=
d
to hash the string to an integer somehow.
       #     Once that is done the reorder and renumber routines should
handle it from there.

       # JB: The DEBUG statements should (perhaps) have a different name.  =
I
use them to track what the program
       #    is doing even when it is working correctly.  Maybe a verbose
switch would be a good replacement here.
       #    the debug statements are called even when there are no eception=
s
thrown (and therefore no tracebacks).

Ok
--=20
Dr. John B. Cole, Research Geneticist
Animal Improvement Programs Laboratory
10300 Baltimore Avenue
BARC-West, Building 005, Room 306
Beltsville, Maryland 20705-2350