Mercurial > repos > mheinzl > variant_analyzer2

changeset 46:f733c425b804 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/Single-Molecule-Genetics/VariantAnalyzerGalaxy/tree/master/tools/variant_analyzer commit ee4a8e6cf290e6c8a4d55f9cd2839d60ab3b11c8
author mheinzl
date Tue, 09 Mar 2021 12:43:22 +0000
parents abb937211f2e
children edf8596463a8
files read2mut.py
diffstat 1 files changed, 26 insertions(+), 23 deletions(-) [+]
line wrap: on
line diff
--- a/read2mut.py	Tue Mar 02 19:09:15 2021 +0000
+++ b/read2mut.py	Tue Mar 09 12:43:22 2021 +0000
@@ -295,7 +295,7 @@
     counter_tier24 = 0
     counter_tier31 = 0
     counter_tier32 = 0
-    counter_tier33 = 0
+    counter_tier25 = 0
     counter_tier4 = 0
     # if chimera_correction:
     #    counter_tier43 = 0
@@ -324,7 +324,7 @@
 
             tier_dict[key1] = {}
             values_tier_dict = [("tier 1.1", 0), ("tier 1.2", 0), ("tier 2.1", 0), ("tier 2.2", 0), ("tier 2.3", 0), ("tier 2.4", 0), ("tier 3.1", 0),
-                                ("tier 3.2", 0), ("tier 3.3", 0), ("tier 4", 0), ("tier 5.1", 0), ("tier 5.2", 0), ("tier 5.3", 0), ("tier 5.4", 0), ("tier 5.5", 0),
+                                ("tier 3.2", 0), ("tier 2.5", 0), ("tier 4", 0), ("tier 5.1", 0), ("tier 5.2", 0), ("tier 5.3", 0), ("tier 5.4", 0), ("tier 5.5", 0),
                                 ("tier 6", 0), ("tier 7", 0)]
             for k, v in values_tier_dict:
                 tier_dict[key1][k] = v
@@ -827,6 +827,9 @@
                             details1 = (total1, total4, total1new, total4new, ref1, ref4, alt1, alt4, ref1f, ref4f, alt1f, alt4f, na1, na4, lowq1, lowq4, beg1, beg4)
                             details2 = (total2, total3, total2new, total3new, ref2, ref3, alt2, alt3, ref2f, ref3f, alt2f, alt3f, na2, na3, lowq2, lowq3, beg2, beg3)
 
+
+                        sum_highTiers = sum([tier_dict[key1][ij] for ij in tier_dict[key1].keys()[:6]])
+
                         # assign tiers
                         if ((all(int(ij) >= 3 for ij in [total1new, total4new]) &
                              all(float(ij) >= 0.75 for ij in [alt1ff, alt4ff])) |
@@ -895,10 +898,10 @@
                             counter_tier32 += 1
                             tier_dict[key1]["tier 3.2"] += 1
 
-                        elif (trimmed) and (len(pure_tags_dict_short[key1]) > 1):
-                            tier = "3.3"
-                            counter_tier33 += 1
-                            tier_dict[key1]["tier 3.3"] += 1
+                        elif (trimmed) and (sum_highTiers > 1):
+                            tier = "2.5"
+                            counter_tier25 += 1
+                            tier_dict[key1]["tier 2.5"] += 1
 
                         elif (trimmed):
                             tier = "4"
@@ -1038,7 +1041,7 @@
                                                 'multi_range': 'L{}:M{} T{}:U{} B{}'.format(row + 1, row + 2, row + 1, row + 2, row + 1, row + 2)})
                         ws1.conditional_format('L{}:M{}'.format(row + 1, row + 2),
                                                {'type': 'formula',
-                                                'criteria': '=OR($B${}="2.1", $B${}="2.2", $B${}="2.3", $B${}="2.4")'.format(row + 1, row + 1, row + 1, row + 1),
+                                                'criteria': '=OR($B${}="2.1", $B${}="2.2", $B${}="2.3", $B${}="2.4", $B${}="2.5")'.format(row + 1, row + 1, row + 1, row + 1, row + 1),
                                                 'format': format3,
                                                 'multi_range': 'L{}:M{} T{}:U{} B{}'.format(row + 1, row + 2, row + 1, row + 2, row + 1, row + 2)})
                         ws1.conditional_format('L{}:M{}'.format(row + 1, row + 2),
@@ -1063,13 +1066,13 @@
     # sheet 2
     if chimera_correction:
         header_line2 = ('variant ID', 'cvrg', 'AC alt (all tiers)', 'AF (all tiers)', 'chimeras in AC alt (all tiers)', 'chimera-corrected cvrg', 'chimera-corrected AF (all tiers)', 'cvrg (tiers 1.1-2.4)', 'AC alt (tiers 1.1-2.4)', 'AF (tiers 1.1-2.4)', 'chimeras in AC alt (tiers 1.1-2.4)', 'chimera-corrected cvrg (tiers 1.1-2.4)', 'chimera-corrected AF (tiers 1.1-2.4)', 'AC alt (orginal DCS)', 'AF (original DCS)',
-                    'tier 1.1', 'tier 1.2', 'tier 2.1', 'tier 2.2', 'tier 2.3', 'tier 2.4',
-                    'tier 3.1', 'tier 3.2', 'tier 3.3', 'tier 4', 'tier 5.1', 'tier 5.2', 'tier 5.3', 'tier 5.4', 'tier 5.5', 'tier 6', 'tier 7', 'AF 1.1-1.2', 'AF 1.1-2.1', 'AF 1.1-2.2',
+                    'tier 1.1', 'tier 1.2', 'tier 2.1', 'tier 2.2', 'tier 2.3', 'tier 2.4', 'tier 2.5',
+                    'tier 3.1', 'tier 3.2', 'tier 4', 'tier 5.1', 'tier 5.2', 'tier 5.3', 'tier 5.4', 'tier 5.5', 'tier 6', 'tier 7', 'AF 1.1-1.2', 'AF 1.1-2.1', 'AF 1.1-2.2',
                     'AF 1.1-2.3', 'AF 1.1-2.4', 'AF 1.1-3.1', 'AF 1.1-3.2', 'AF 1.1-4.1', 'AF 1.1-4.2', 'AF 1.1-5.1', 'AF 1.1-5.2', 'AF 1.1-5.3', 'AF 1.1-5.4', 'AF 1.1-5.5', 'AF 1.1-6')
     else:
         header_line2 = ('variant ID', 'cvrg', 'AC alt (all tiers)', 'AF (all tiers)', 'cvrg (tiers 1.1-2.4)', 'AC alt (tiers 1.1-2.4)', 'AF (tiers 1.1-2.4)', 'AC alt (orginal DCS)', 'AF (original DCS)',
-                        'tier 1.1', 'tier 1.2', 'tier 2.1', 'tier 2.2', 'tier 2.3', 'tier 2.4',
-                        'tier 3.1', 'tier 3.2', 'tier 3.3', 'tier 4', 'tier 5.1', 'tier 5.2', 'tier 5.3', 'tier 5.4', 'tier 5.5', 'tier 6', 'tier 7', 'AF 1.1-1.2', 'AF 1.1-2.1', 'AF 1.1-2.2',
+                        'tier 1.1', 'tier 1.2', 'tier 2.1', 'tier 2.2', 'tier 2.3', 'tier 2.4', 'tier 2.5',
+                        'tier 3.1', 'tier 3.2', 'tier 4', 'tier 5.1', 'tier 5.2', 'tier 5.3', 'tier 5.4', 'tier 5.5', 'tier 6', 'tier 7', 'AF 1.1-1.2', 'AF 1.1-2.1', 'AF 1.1-2.2',
                         'AF 1.1-2.3', 'AF 1.1-2.4', 'AF 1.1-3.1', 'AF 1.1-3.2', 'AF 1.1-4.1', 'AF 1.1-4.2', 'AF 1.1-5.1', 'AF 1.1-5.2', 'AF 1.1-5.3', 'AF 1.1-5.4', 'AF 1.1-5.5', 'AF 1.1-6')
 
     ws2.write_row(0, 0, header_line2)
@@ -1107,14 +1110,14 @@
                     fraction_chimeras = 0.
                 new_cvrg = cvrg * (1. - fraction_chimeras)
                 lst.extend([chimeras_all, new_cvrg, safe_div(new_alt, new_cvrg)])
-            lst.extend([(cvrg - sum(used_tiers[-11:])), sum(used_tiers[0:6]), safe_div(sum(used_tiers[0:6]), (cvrg - sum(used_tiers[-11:])))])
+            lst.extend([(cvrg - sum(used_tiers[-10:])), sum(used_tiers[0:7]), safe_div(sum(used_tiers[0:7]), (cvrg - sum(used_tiers[-10:])))])
             if chimera_correction:
                 chimeras_all = chimera_dict[key1][1]
-                new_alt = sum(used_tiers[0:6]) - chimeras_all
-                fraction_chimeras = safe_div(chimeras_all, float(sum(used_tiers[0:6])))
+                new_alt = sum(used_tiers[0:7]) - chimeras_all
+                fraction_chimeras = safe_div(chimeras_all, float(sum(used_tiers[0:7])))
                 if fraction_chimeras is None:
                     fraction_chimeras = 0.
-                new_cvrg = (cvrg - sum(used_tiers[-11:])) * (1. - fraction_chimeras)
+                new_cvrg = (cvrg - sum(used_tiers[-10:])) * (1. - fraction_chimeras)
                 lst.extend([chimeras_all, new_cvrg, safe_div(new_alt, new_cvrg)])
             lst.extend([alt_count, safe_div(alt_count, cvrg)])
             lst.extend(used_tiers)
@@ -1123,18 +1126,18 @@
             ws2.write_row(row + 1, 0, lst)
             if chimera_correction:
                 ws2.conditional_format('P{}:Q{}'.format(row + 2, row + 2), {'type': 'formula', 'criteria': '=$P$1="tier 1.1"', 'format': format12, 'multi_range': 'P{}:Q{} P1:Q1'.format(row + 2, row + 2)})
-                ws2.conditional_format('R{}:U{}'.format(row + 2, row + 2), {'type': 'formula', 'criteria': '=$R$1="tier 2.1"', 'format': format32, 'multi_range': 'R{}:U{} R1:U1'.format(row + 2, row + 2)})
-                ws2.conditional_format('V{}:AF{}'.format(row + 2, row + 2), {'type': 'formula', 'criteria': '=$V$1="tier 3.1"', 'format': format22, 'multi_range': 'V{}:AF{} V1:AF1'.format(row + 2, row + 2)})
+                ws2.conditional_format('R{}:V{}'.format(row + 2, row + 2), {'type': 'formula', 'criteria': '=$R$1="tier 2.1"', 'format': format32, 'multi_range': 'R{}:V{} R1:V1'.format(row + 2, row + 2)})
+                ws2.conditional_format('W{}:AF{}'.format(row + 2, row + 2), {'type': 'formula', 'criteria': '=$W$1="tier 3.1"', 'format': format22, 'multi_range': 'W{}:AF{} W1:AF1'.format(row + 2, row + 2)})
             else:
                 ws2.conditional_format('J{}:K{}'.format(row + 2, row + 2), {'type': 'formula', 'criteria': '=$J$1="tier 1.1"', 'format': format12, 'multi_range': 'J{}:K{} J1:K1'.format(row + 2, row + 2)})
-                ws2.conditional_format('L{}:O{}'.format(row + 2, row + 2), {'type': 'formula', 'criteria': '=$L$1="tier 2.1"', 'format': format32, 'multi_range': 'L{}:O{} L1:O1'.format(row + 2, row + 2)})
-                ws2.conditional_format('P{}:Z{}'.format(row + 2, row + 2), {'type': 'formula', 'criteria': '=$P$1="tier 3.1"', 'format': format22, 'multi_range': 'P{}:Z{} P1:Z1'.format(row + 2, row + 2)})
+                ws2.conditional_format('L{}:P{}'.format(row + 2, row + 2), {'type': 'formula', 'criteria': '=$L$1="tier 2.1"', 'format': format32, 'multi_range': 'L{}:P{} L1:P1'.format(row + 2, row + 2)})
+                ws2.conditional_format('Q{}:Z{}'.format(row + 2, row + 2), {'type': 'formula', 'criteria': '=$P$1="tier 3.1"', 'format': format22, 'multi_range': 'Q{}:Z{} Q1:Z1'.format(row + 2, row + 2)})
             row += 1
 
     # sheet 3
     sheet3 = [("tier 1.1", counter_tier11), ("tier 1.2", counter_tier12), ("tier 2.1", counter_tier21),
-              ("tier 2.2", counter_tier22), ("tier 2.3", counter_tier23), ("tier 2.4", counter_tier24),
-              ("tier 3.1", counter_tier31), ("tier 3.2", counter_tier32), ("tier 3.3", counter_tier33), ("tier 4", counter_tier4),
+              ("tier 2.2", counter_tier22), ("tier 2.3", counter_tier23), ("tier 2.4", counter_tier24), ("tier 2.5", counter_tier25), 
+              ("tier 3.1", counter_tier31), ("tier 3.2", counter_tier32), ("tier 4", counter_tier4),
               ("tier 5.1", counter_tier51), ("tier 5.2", counter_tier52),
               ("tier 5.3", counter_tier53), ("tier 5.4", counter_tier54), ("tier 5.5", counter_tier55), ("tier 6", counter_tier6), ("tier 7", counter_tier7)]
 
@@ -1149,7 +1152,7 @@
                                 'format': format1})
         ws3.conditional_format('A{}:B{}'.format(i + 2, i + 2),
                                {'type': 'formula',
-                                'criteria': '=OR($A${}="tier 2.1", $A${}="tier 2.2", $A${}="tier 2.3", $A${}="tier 2.4")'.format(i + 2, i + 2, i + 2, i + 2),
+                                'criteria': '=OR($A${}="tier 2.1", $A${}="tier 2.2", $A${}="tier 2.3", $A${}="tier 2.4", $A${}="tier 2.5")'.format(i + 2, i + 2, i + 2, i + 2, i + 2),
                                 'format': format3})
         ws3.conditional_format('A{}:B{}'.format(i + 2, i + 2),
                                {'type': 'formula',
@@ -1255,7 +1258,7 @@
             ws3.write_row(start_row + 2 + row + i + k + 2, 0, ex[k])
         ws3.conditional_format('L{}:M{}'.format(start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 3), {'type': 'formula', 'criteria': '=OR($B${}="1.1", $B${}="1.2")'.format(start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 2), 'format': format13, 'multi_range': 'L{}:M{} T{}:U{} B{}'.format(start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 3, start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 3, start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 3)})
         ws3.conditional_format('L{}:M{}'.format(start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 3),
-                               {'type': 'formula', 'criteria': '=OR($B${}="2.1",$B${}="2.2", $B${}="2.3", $B${}="2.4")'.format(start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 2),
+                               {'type': 'formula', 'criteria': '=OR($B${}="2.1",$B${}="2.2", $B${}="2.3", $B${}="2.4", $B${}="2.5")'.format(start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 2),
                                 'format': format33,
                                 'multi_range': 'L{}:M{} T{}:U{} B{}'.format(start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 3, start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 3, start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 3)})
         ws3.conditional_format('L{}:M{}'.format(start_row + 2 + row + i + k + 2, start_row + 2 + row + i + k + 3),