From e5a69ed19772c8cd1208c1ded16872947d6e34a3 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Dernat=20R=C3=A9my?= <remy.dernat@umontpellier.fr>
Date: Fri, 22 Nov 2019 11:22:41 +0100
Subject: [PATCH] adding comments
---
orthofinder/orthofinder.py | 8 ++++++--
orthofinder/scripts/blast_file_processor.py | 2 +-
2 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/orthofinder/orthofinder.py b/orthofinder/orthofinder.py
index 2fd34fe..a05a013 100755
--- a/orthofinder/orthofinder.py
+++ b/orthofinder/orthofinder.py
@@ -298,10 +298,12 @@ scnorm
class scnorm:
@staticmethod
def loglinear(x, a, b):
- return a*np.log10(x)+b
+ # return the loglinear function
+ return a*np.log10(x)+b
@staticmethod
def GetLengthArraysForMatrix(m, len_i, len_j):
+ # return (I,j) positions non zeros values for the m matrix
I, J = m.nonzero()
scores = [v for row in m.data for v in row] # use fact that it's lil
Li = np.array(len_i[I])
@@ -336,15 +338,17 @@ class scnorm:
@staticmethod
def CalculateFittingParameters(Lf, S):
+ # curve_fit doc.: https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.curve_fit.html
pars,covar = curve_fit(scnorm.loglinear, Lf, np.log10(S))
return pars
- @staticmethod
+ @staticmethod
def NormaliseScoresByLogLengthProduct(b, Lq, Lh, params):
rangeq = list(range(len(Lq)))
rangeh = list(range(len(Lh)))
li_vals = Lq**(-params[0])
lj_vals = Lh**(-params[0])
+ # l{i,j}_matrix are built using l{i,j}_vals data using (range{q,h},range{q,h}) positions
li_matrix = sparse.csr_matrix((li_vals, (rangeq, rangeq)))
lj_matrix = sparse.csr_matrix((lj_vals, (rangeh, rangeh)))
return sparse.lil_matrix(10**(-params[1]) * li_matrix * b * lj_matrix)
diff --git a/orthofinder/scripts/blast_file_processor.py b/orthofinder/scripts/blast_file_processor.py
index f882c1a..232b2f7 100644
--- a/orthofinder/scripts/blast_file_processor.py
+++ b/orthofinder/scripts/blast_file_processor.py
@@ -104,7 +104,7 @@ def GetBLAST6Scores(seqsInfo, blastDir_list, iSpecies, jSpecies, qExcludeSelfHit
return B
def WriteNormalizedBlastFile(blastDir_list, B, iSpecies, jSpecies, sep = "_", qDoubleBlast=True):
- qSameSpecies = iSpecies==jSpecies
+ # checking species order to get the right filenames
if not qDoubleBlast:
qRev = (iSpecies > jSpecies)
else:
--
GitLab