Commit 6f0ebead authored by Martin Larralde's avatar Martin Larralde
Browse files

Add support for Bayes discriminant distance in `Peptide.structural_class`

parent 19e6d745
......@@ -8,6 +8,11 @@ import random
import statistics
import typing
from numpy import prod
except ImportError:
from ._npcompat import prod
from . import tables, datasets
__all__ = ["Peptide", "tables", "datasets"]
......@@ -1282,7 +1287,7 @@ class Peptide(typing.Sequence[str]):
centroids. Use `"Chou"` to load the frequencies of the 64
proteins analyzed in Chou (1989), `"Nakashima"` to use
the normalized frequencies of the 135 proteins analyzed in
Nakashima *et al* (1986) and Zhang & Chou (1995), or
Nakashima *et al.* (1986) and Zhang & Chou (1995), or
`"ChouZhang"` to load the frequencies of 120 proteins used
in Chou & Zhang (1995).
distance (`str`): The distance metric to use in the 20-D space
......@@ -1291,8 +1296,10 @@ class Peptide(typing.Sequence[str]):
the Manhattan distance like in Chou (1989), `"euclidean"`
to use the Euclidean distance like in Nakashima *et al*
(1986), `"correlation"` to use the correlation distance
like in Chou & Zhang (1992), or `"mahalanobis"` to use
the Mahalanobis distance like in Chou & Zhang (1995).
like in Chou & Zhang (1992), `"mahalanobis"` to use
the Mahalanobis distance like in Chou & Zhang (1995),
or `"discriminant"` to use the Bayes discriminant like in
Chou *et al.* (1998).
`str`: The structural class the protein most likely belongs to.
......@@ -1353,6 +1360,10 @@ class Peptide(typing.Sequence[str]):
- Chou, K-C., W-M. Liu, G. M. Maggiora, and C-T. Zhang.
*Prediction and Classification of Domain Structural Classes*.
Proteins: Structure, Function, and Genetics.
Apr 1998;31(1):97–103. PMID:9552161.
- Chou, K-C., and C-T. Zhang.
*Prediction of Protein Structural Classes*. Critical Reviews
in Biochemistry and Molecular Biology. Feb 1995;30:275–349.
......@@ -1428,7 +1439,7 @@ class Peptide(typing.Sequence[str]):
table = tables["mean"]
s = sum((pep_frequencies[x]-table[x])**2 for x in table)
distances[name] = math.sqrt(s)
elif distance == "mahalanobis":
elif distance == "mahalanobis" or distance == "discriminant":
x = [pep_frequencies[aa]*100 for aa in sorted(self._CODE1[:20])]
for name,tables in dataset.items():
if name == "all":
......@@ -1447,6 +1458,10 @@ class Peptide(typing.Sequence[str]):
for i in range(20)
distances[name] = sum(y[i]**2 / eivals[i] for i in range(1, 20))
# if Bayes discriminant is requested, add the logarithm of
# the product of all non-null eigenvalues
if distance == "discriminant":
distances[name] += math.log(prod(einvals[1:]))
if not distances:
raise ValueError(
f"Cannot use {frequencies!r} frequencies with "
......@@ -1456,7 +1471,6 @@ class Peptide(typing.Sequence[str]):
raise ValueError(f"Invalid distance: {distance!r}")
# find the most likely structural class based on the distance
# print(distances)
return min(distances, key=distances.get)
# --- Descriptors --------------------------------------------------------
import functools
import operator
def prod(a):
"""Return the product of an iterable of numbers
return functools.reduce(, a)
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment