camelot-frs 0.0.2

A frame representation system in Python.

Camelot: A Pythonic Frame Representation System

Camelot is a Frame Representation System (FRS) for Python. It is inspired by the Ocelot FRS which powers the Pathway Tools software from SRI International's Bioinformatcs Research Group. Thus, it is called 'Camelot' as a Pythonic Ocelot.

Current Capabilities

• Slot annotations

• Transitive Closure for is-a (i.e., subsumption) and part-of relationships

• Supports the import of KBs in the attribute-value format of Pathway/Genome Databases as generated by Pathway Tools

• Implementation of select functions from the Pathway Tools API

• Enrichment Analysis

Planned Capabilities

• Visualization of Ontology structure using GraphViz

• Ties to BioPython for handling sequence data

• Support data loaders for GO, ENZYME, NCBI Taxonomy DB, and UniProt

How to Build

Run the following at the top of the package repository:

make clean-package
make create-package

How to Install

Camelot FRS is released as a Python package. It can easily be installed using FIXME.

To manually install the package from the Git repository, do the following at the top-level of the repository:

make install-package

For development purposes, the following will install an 'editable' version of the package:

make install-dev-package

A Simple Tutorial

## Import Camelot functions for working with ontology:
from camelot_frs.pgdb_loader import load_pgdb
from camelot_frs.pgdb_api    import reactions_of_compound
from camelot_frs.camelot_frs import get_kb, get_frame, get_frames, get_frame_all_parents, get_frame_all_children

## Load MetaCyc:
load_pgdb("/home/ubuntu/data/metacyc/23.1/data")

## Assign the MetaCyc Knowledge Base (KB) to a variable:
metaKB = get_kb('META')

### Find all reactions of 'ACET':

## Get the frame for acetic acid:
acet = get_frame(metaKB, 'ACET')

## Get all reactions that directly 

## Get all reactions that indirectly consume acetic acid:
## (Either an instance or sub-class or a super-class of acetic acid is consumed in the reaction)
indirect_rxns = reactions_of_compound(acet, generic_rxns_p=True, consumed_p=True)

A Tutorial using MetaCyc

Design Decisions

From my years of experience working with the Ocelot DB, I had some definite opinions about how to design the user experience with Camelot. The following is a summary of some of these decisions:

• All KBs and frames are objects
• There are no symbol-equivalent representations of KBs or frames
• Since a user has to deal with objects directly, objects have a rational and human-readable display form, such as 'frame:proton' for a proton object, rather than '(Python object 0x399823498)'
• Since all frames are represented as objects, each object points specifically to the KB that it comes from (and every KB has a link to the frames it contains). Only frame look-up functions require the KB as a co-argument. This greatly simplifies the code.
• KB-agnostic code is kept strictly separate from KB-specific code (e.g., the PGDB Loader)
• Core algorithms for processing ontology datastructures are separated out from the 'business logic'
• Instead of having separate functions for dealing with classes versus instances, functions are written to be as generic as possible, and named arguments are optional for specifying between them (e.g., get_frame_all_children can specify optionally to return only classes, only instances, or both, which is the default)
• All slots are lists, even if they usually only have a single value
• Accessing a non-existent slot throws an error. Schema violations should cause code to fail and fail early