15 Nov 00 Brad

* Added SubsMat tests, sample files and expected outputs.
* Updated alignment test to use SubsMat.FreqTable.

Tests/SubsMat/aaDistrib.txt

@@ -0,0 +1,20 @@
+A 2377 0.0852491
+C 383 0.013736
+D 1669 0.0598573
+E 1655 0.0593552
+F 1167 0.0418535
+G 2185 0.0783632
+H 627 0.0224868
+I 1455 0.0521823
+K 1614 0.0578847
+L 2360 0.0846394
+M 651 0.0233476
+N 1387 0.0497436
+P 1191 0.0427142
+Q 1022 0.0366532
+R 1282 0.0459778
+S 1664 0.0596779
+T 1599 0.0573468
+V 2052 0.0735932
+W 421 0.0150988
+Y 1119 0.040132

Tests/SubsMat/protein_count.txt

@@ -0,0 +1,20 @@
+A 2377
+C 383
+D 1669
+E 1655
+F 1167
+G 2185
+H 627
+I 1455
+K 1614
+L 2360
+M 651
+N 1387
+P 1191
+Q 1022
+R 1282
+S 1664
+T 1599
+V 2052
+W 421
+Y 1119

Tests/SubsMat/protein_freq.txt

@@ -0,0 +1,20 @@
+A 0.0852491
+C 0.013736
+D 0.0598573
+E 0.0593552
+F 0.0418535
+G 0.0783632
+H 0.0224868
+I 0.0521823
+K 0.0578847
+L 0.0846394
+M 0.0233476
+N 0.0497436
+P 0.0427142
+Q 0.0366532
+R 0.0459778
+S 0.0596779
+T 0.0573468
+V 0.0735932
+W 0.0150988
+Y 0.040132

Tests/SubsMat/redAADistrib.txt

@@ -0,0 +1,7 @@
+F 9278 0.332748
+R 3334 0.119571
+O 5672 0.203422
+T 2896 0.103862
+N 3324 0.119212
+G 2185 0.0783632
+P 1191 0.0427142

Tests/SubsMat/redAltAADistrib.txt

@@ -0,0 +1,8 @@
+A 3411 0.2104153
+F 5867 0.1223327
+R 3334 0.119571
+O 5672 0.203422
+T 2896 0.103862
+N 3324 0.119212
+G 2185 0.0783632
+P 1191 0.0427142

Tests/test_SubsMat.py

@@ -0,0 +1,40 @@
+import cPickle
+import sys
+from Bio.SubsMat import *
+
+f = sys.stdout
+ftab_prot = FreqTable.read_count(open('SubsMat/protein_count.txt'))
+ctab_prot = FreqTable.read_freq(open('SubsMat/protein_freq.txt'))
+f.write("Check differences between derived and true frequencies for each\n")
+f.write("letter. Differences should be very small\n")
+for i in ftab_prot.alphabet.letters:
+    f.write("%s %f\n" % (i, abs(ftab_prot[i] - ctab_prot[i])))
+    
+acc_rep_mat = cPickle.load(open('SubsMat/acc_rep_mat.pik'))
+acc_rep_mat = SubsMat.SeqMat(acc_rep_mat)
+obs_freq_mat = SubsMat._build_obs_freq_mat(acc_rep_mat)
+ftab_prot2 = SubsMat._exp_freq_table_from_obs_freq(obs_freq_mat)
+obs_freq_mat.print_mat(f=f,format=" %4.3f")
+
+
+f.write("Diff between supplied and matrix-derived frequencies, should be small\n")
+for i in ftab_prot.keys():
+    f.write("%s %.2f\n" % (i,abs(ftab_prot[i] - ftab_prot2[i])))
+
+s = 0.
+f.write("Calculating sum of letters for an observed frequency matrix\n")
+obs_freq_mat.all_letters_sum()
+for i in obs_freq_mat.sum_letters.keys():
+    f.write("%s\t%.2f\n" % (i, obs_freq_mat.sum_letters[i]))
+    s += obs_freq_mat.sum_letters[i]
+f.write("Total sum %.2f should be 1.0\n" % (s))
+lo_mat_prot = \
+SubsMat.make_log_odds_matrix(acc_rep_mat=acc_rep_mat,round_digit=1) #,ftab_prot
+f.write("\nLog odds matrix\n")
+lo_mat_prot.print_mat(f=f,format=" %.2f",alphabet='AVILMCFWYHSTNQKRDEGP')
+
+f.write("\nTesting MatrixInfo\n")
+for i in MatrixInfo.available_matrices:
+    mat = SubsMat.SeqMat(getattr(MatrixInfo,i))
+    f.write("\n%s\n------------\n" % i)
+    mat.print_mat(f=f)

Tests/test_align.py

@@ -12,10 +12,12 @@ o Converting between formats"""
 import os 
 
 # biopython
+from Bio.Alphabet import IUPAC
 from Bio.Clustalw import Clustalw
 from Bio.Align.FormatConvert import FormatConverter
 from Bio.Align import AlignInfo
 from Bio.Fasta import FastaAlign
+from Bio.SubsMat import FreqTable
 
 print "testing reading and writing clustal format..."
 test_dir = os.path.join(os.getcwd(), 'Clustalw')
@@ -74,8 +76,11 @@ e_freq = {'G' : 0.25,
           'A' : 0.25,
           'T' : 0.25}
 
+e_freq_table = FreqTable.FreqTable(e_freq, FreqTable.FREQ,
+                                   IUPAC.unambiguous_dna)
+
 print 'relative information:', align_info.information_content(
-                                   expected_freqs = e_freq,
+                                   e_freq_table = e_freq_table,
                                    chars_to_ignore = ['N'])
 
 print "testing reading and writing fasta format..."