From 355d219f29111201b611546107fc8de5b16d0a77 Mon Sep 17 00:00:00 2001
From: Vitaliy Mysak <vmysak@bccfe.ca>
Date: Mon, 18 Sep 2023 11:43:02 -0700
Subject: [PATCH] add qlen to holistic info

---
 intact/intact.py                              |  2 +
 tests/expected-results-large-csv/holistic.csv | 84 +++++++++----------
 tests/expected-results-large/holistic.json    | 41 +++++++++
 .../expected-results-single-csv/holistic.csv  |  4 +-
 tests/expected-results-single/holistic.json   |  1 +
 tests/expected-results-small-csv/holistic.csv | 12 +--
 tests/expected-results-small/holistic.json    |  5 ++
 7 files changed, 99 insertions(+), 50 deletions(-)

diff --git a/intact/intact.py b/intact/intact.py
index 790e67c..044ad11 100644
--- a/intact/intact.py
+++ b/intact/intact.py
@@ -100,6 +100,7 @@ class FoundORF:
 
 @dataclass
 class HolisticInfo:
+    qlen: int = dataclasses.field(default=None)
     hypermutation_probablility: float = dataclasses.field(default=None)
     inferred_subtype: str  = dataclasses.field(default=None)
     blast_matched_qlen: int  = dataclasses.field(default=None) # number of query nucleotides matched to a known reference sequence
@@ -854,6 +855,7 @@ def intact( working_dir,
 
             sequence_errors = []
             holistic = HolisticInfo()
+            holistic.qlen = len(sequence)
             holistic.inferred_subtype = reference_name
             holistic.blast_n_conseqs = len(blast_rows)
 
diff --git a/tests/expected-results-large-csv/holistic.csv b/tests/expected-results-large-csv/holistic.csv
index 81472cb..383c944 100644
--- a/tests/expected-results-large-csv/holistic.csv
+++ b/tests/expected-results-large-csv/holistic.csv
@@ -1,42 +1,42 @@
-seqid,hypermutation_probablility,inferred_subtype,blast_matched_qlen,blast_sseq_coverage,blast_qseq_coverage,blast_n_conseqs
-KX505501.1,0.7087072014754221,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,1997,0.2498199403230785,1.2158237356034052,4
-MN691959,0.19593905853945925,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9493,1.0816956477003807,1.1086063415148004,3
-MN692074,0.36378645339477633,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,4178,0.5041670953801831,1.1728099569171853,4
-MN692145,0.1661041079701131,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9689,1.1304660973351168,1.1271545051088863,3
-MN090335,0.1754017863888554,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9069,0.9842576396748637,1.0603153600176425,3
-MN090376,0.026007919521734202,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,8985,0.9784957300133759,1.0604340567612687,3
-MK115581.1,0.6897199265079494,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9495,0.9870357032616525,1.0046340179041602,2
-MK115690.1,0.05065930954004094,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9689,0.988167506945159,0.9949427185468056,2
-MK115571.1,0.8012585672082311,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9394,0.9866241382858318,1.0113902490951672,2
-MK115514.1,0.6458974386368621,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9382,0.9864183557979216,1.0173736943082499,2
-MK115488.1,0.6511896911074662,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9623,1.0255170285008746,1.0325262392185388,6
-MK115030.1,0.031598631869680704,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9126,0.9940323078506019,1.0655270655270654,3
-MK115498.1,0.8339748776671196,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9461,0.9866241382858318,1.0080329774865235,2
-MK115211.1,0.11689558806708,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9032,0.9950612202901533,1.0598981399468557,3
-MK115158.1,0.002572269807584293,Ref.47_BF.ES.08.P1942.GQ372987,9143,0.9921577414295317,0.9699223449633599,1
-MK114705.1,0.14449377496074622,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9411,1.0980553554892478,1.122622463075125,6
-MK114856.1,1.0,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9477,1.0485646671468258,1.0812493405085997,4
-MK115009.1,1.0,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9207,0.9965016977055252,1.0590854784403172,3
-MK115387.1,0.5412311092694289,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9136,0.9796275336968824,1.040936952714536,2
-MK115491.1,0.8951015182445495,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9422,1.003704084782385,1.0299299511780937,2
-MK116110.1,0.07021438897893317,Ref.B.TH.90.BK132.AY173951,8967,0.9957759004001778,0.9972119995539199,3
-MK115527.1,0.7689834393883834,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9481,0.986727029529787,1.0056956017297753,2
-MK114997.1,0.054959132555391754,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9055,0.9784957300133759,1.0516841524019878,2
-MK115518.1,0.6385326595592609,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9537,0.9847720958946393,0.9996854356715948,3
-MK115065.1,0.033517722768753644,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9214,1.0080255170285009,1.069459518124593,6
-MK115464.1,1.0,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9663,0.9838460746990431,0.9893407844354756,2
-MK115530.1,0.5789377103398377,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9544,0.9866241382858318,0.9992665549036044,2
-MK115520.1,0.5200353682902832,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9589,0.9787015125012861,0.987902805297737,3
-MK115503.1,0.4263025132504157,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9617,0.9870357032616525,0.9953207861079338,2
-MK115570.1,0.738578434638724,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9485,0.9866241382858318,1.0057986294148655,2
-MK115509.1,0.7866198309713798,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9353,0.9866241382858318,1.0197797498128942,2
-MK115702.1,0.14401391767451666,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9098,0.987447268237473,1.0596834469114091,4
-MK115095.1,1.0,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9137,0.9907397880440375,1.060085367188355,2
-MK115490.1,0.8863248655310947,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9347,0.9848749871385946,1.0204343639670483,3
-MK115576.1,0.818189227062389,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9266,0.9917687004835889,1.0342110943233327,3
-OQ092466,0.3876036547663967,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9686,1.1015536577837226,1.1192442700805285,3
-OQ092463,0.21628713708846803,Ref.B.TH.90.BK132.AY173951,9605,1.0529124055135617,0.9884435190005205,2
-OQ092465,0.02412789935966586,Ref.28_BF.BR.99.BREPM12817.DQ085874,9659,1.0429696287964005,0.9620043482762191,2
-OQ092462,0.10777665573070194,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9714,1.130671879823027,1.1301214741610048,3
-OQ092464,0.006887768010151674,Ref.28_BF.BR.99.BREPM12817.DQ085874,9556,1.0427446569178853,0.9678735872750105,2
-OQ092467,0.6416537859942263,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9936,1.1308776623109373,1.0962157809983897,3
+seqid,qlen,hypermutation_probablility,inferred_subtype,blast_matched_qlen,blast_sseq_coverage,blast_qseq_coverage,blast_n_conseqs
+KX505501.1,1997,0.7087072014754221,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,1997,0.2498199403230785,1.2158237356034052,4
+MN691959,9493,0.19593905853945925,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9493,1.0816956477003807,1.1086063415148004,3
+MN692074,4178,0.36378645339477633,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,4178,0.5041670953801831,1.1728099569171853,4
+MN692145,9689,0.1661041079701131,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9689,1.1304660973351168,1.1271545051088863,3
+MN090335,9069,0.1754017863888554,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9069,0.9842576396748637,1.0603153600176425,3
+MN090376,8985,0.026007919521734202,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,8985,0.9784957300133759,1.0604340567612687,3
+MK115581.1,9495,0.6897199265079494,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9495,0.9870357032616525,1.0046340179041602,2
+MK115690.1,9689,0.05065930954004094,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9689,0.988167506945159,0.9949427185468056,2
+MK115571.1,9394,0.8012585672082311,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9394,0.9866241382858318,1.0113902490951672,2
+MK115514.1,9382,0.6458974386368621,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9382,0.9864183557979216,1.0173736943082499,2
+MK115488.1,9623,0.6511896911074662,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9623,1.0255170285008746,1.0325262392185388,6
+MK115030.1,9126,0.031598631869680704,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9126,0.9940323078506019,1.0655270655270654,3
+MK115498.1,9461,0.8339748776671196,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9461,0.9866241382858318,1.0080329774865235,2
+MK115211.1,9032,0.11689558806708,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9032,0.9950612202901533,1.0598981399468557,3
+MK115158.1,9143,0.002572269807584293,Ref.47_BF.ES.08.P1942.GQ372987,9143,0.9921577414295317,0.9699223449633599,1
+MK114705.1,9411,0.14449377496074622,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9411,1.0980553554892478,1.122622463075125,6
+MK114856.1,9477,1.0,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9477,1.0485646671468258,1.0812493405085997,4
+MK115009.1,9207,1.0,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9207,0.9965016977055252,1.0590854784403172,3
+MK115387.1,9136,0.5412311092694289,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9136,0.9796275336968824,1.040936952714536,2
+MK115491.1,9422,0.8951015182445495,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9422,1.003704084782385,1.0299299511780937,2
+MK116110.1,8967,0.07021438897893317,Ref.B.TH.90.BK132.AY173951,8967,0.9957759004001778,0.9972119995539199,3
+MK115527.1,9481,0.7689834393883834,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9481,0.986727029529787,1.0056956017297753,2
+MK114997.1,9055,0.054959132555391754,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9055,0.9784957300133759,1.0516841524019878,2
+MK115518.1,9537,0.6385326595592609,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9537,0.9847720958946393,0.9996854356715948,3
+MK115065.1,9214,0.033517722768753644,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9214,1.0080255170285009,1.069459518124593,6
+MK115464.1,9663,1.0,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9663,0.9838460746990431,0.9893407844354756,2
+MK115530.1,9544,0.5789377103398377,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9544,0.9866241382858318,0.9992665549036044,2
+MK115520.1,9589,0.5200353682902832,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9589,0.9787015125012861,0.987902805297737,3
+MK115503.1,9617,0.4263025132504157,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9617,0.9870357032616525,0.9953207861079338,2
+MK115570.1,9485,0.738578434638724,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9485,0.9866241382858318,1.0057986294148655,2
+MK115509.1,9353,0.7866198309713798,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9353,0.9866241382858318,1.0197797498128942,2
+MK115702.1,9098,0.14401391767451666,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9098,0.987447268237473,1.0596834469114091,4
+MK115095.1,9137,1.0,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9137,0.9907397880440375,1.060085367188355,2
+MK115490.1,9347,0.8863248655310947,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9347,0.9848749871385946,1.0204343639670483,3
+MK115576.1,9266,0.818189227062389,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9266,0.9917687004835889,1.0342110943233327,3
+OQ092466,9686,0.3876036547663967,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9686,1.1015536577837226,1.1192442700805285,3
+OQ092463,9605,0.21628713708846803,Ref.B.TH.90.BK132.AY173951,9605,1.0529124055135617,0.9884435190005205,2
+OQ092465,9659,0.02412789935966586,Ref.28_BF.BR.99.BREPM12817.DQ085874,9659,1.0429696287964005,0.9620043482762191,2
+OQ092462,9714,0.10777665573070194,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9714,1.130671879823027,1.1301214741610048,3
+OQ092464,9556,0.006887768010151674,Ref.28_BF.BR.99.BREPM12817.DQ085874,9556,1.0427446569178853,0.9678735872750105,2
+OQ092467,9936,0.6416537859942263,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9936,1.1308776623109373,1.0962157809983897,3
diff --git a/tests/expected-results-large/holistic.json b/tests/expected-results-large/holistic.json
index 9a0e5f8..f98371a 100644
--- a/tests/expected-results-large/holistic.json
+++ b/tests/expected-results-large/holistic.json
@@ -1,5 +1,6 @@
 {
     "KX505501.1": {
+        "qlen": 1997,
         "hypermutation_probablility": 0.7087072014754221,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 1997,
@@ -8,6 +9,7 @@
         "blast_n_conseqs": 4
     },
     "MN691959": {
+        "qlen": 9493,
         "hypermutation_probablility": 0.19593905853945925,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9493,
@@ -16,6 +18,7 @@
         "blast_n_conseqs": 3
     },
     "MN692074": {
+        "qlen": 4178,
         "hypermutation_probablility": 0.36378645339477633,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 4178,
@@ -24,6 +27,7 @@
         "blast_n_conseqs": 4
     },
     "MN692145": {
+        "qlen": 9689,
         "hypermutation_probablility": 0.1661041079701131,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9689,
@@ -32,6 +36,7 @@
         "blast_n_conseqs": 3
     },
     "MN090335": {
+        "qlen": 9069,
         "hypermutation_probablility": 0.1754017863888554,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9069,
@@ -40,6 +45,7 @@
         "blast_n_conseqs": 3
     },
     "MN090376": {
+        "qlen": 8985,
         "hypermutation_probablility": 0.026007919521734202,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 8985,
@@ -48,6 +54,7 @@
         "blast_n_conseqs": 3
     },
     "MK115581.1": {
+        "qlen": 9495,
         "hypermutation_probablility": 0.6897199265079494,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9495,
@@ -56,6 +63,7 @@
         "blast_n_conseqs": 2
     },
     "MK115690.1": {
+        "qlen": 9689,
         "hypermutation_probablility": 0.05065930954004094,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9689,
@@ -64,6 +72,7 @@
         "blast_n_conseqs": 2
     },
     "MK115571.1": {
+        "qlen": 9394,
         "hypermutation_probablility": 0.8012585672082311,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9394,
@@ -72,6 +81,7 @@
         "blast_n_conseqs": 2
     },
     "MK115514.1": {
+        "qlen": 9382,
         "hypermutation_probablility": 0.6458974386368621,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9382,
@@ -80,6 +90,7 @@
         "blast_n_conseqs": 2
     },
     "MK115488.1": {
+        "qlen": 9623,
         "hypermutation_probablility": 0.6511896911074662,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9623,
@@ -88,6 +99,7 @@
         "blast_n_conseqs": 6
     },
     "MK115030.1": {
+        "qlen": 9126,
         "hypermutation_probablility": 0.031598631869680704,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9126,
@@ -96,6 +108,7 @@
         "blast_n_conseqs": 3
     },
     "MK115498.1": {
+        "qlen": 9461,
         "hypermutation_probablility": 0.8339748776671196,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9461,
@@ -104,6 +117,7 @@
         "blast_n_conseqs": 2
     },
     "MK115211.1": {
+        "qlen": 9032,
         "hypermutation_probablility": 0.11689558806708,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9032,
@@ -112,6 +126,7 @@
         "blast_n_conseqs": 3
     },
     "MK115158.1": {
+        "qlen": 9143,
         "hypermutation_probablility": 0.002572269807584293,
         "inferred_subtype": "Ref.47_BF.ES.08.P1942.GQ372987",
         "blast_matched_qlen": 9143,
@@ -120,6 +135,7 @@
         "blast_n_conseqs": 1
     },
     "MK114705.1": {
+        "qlen": 9411,
         "hypermutation_probablility": 0.14449377496074622,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9411,
@@ -128,6 +144,7 @@
         "blast_n_conseqs": 6
     },
     "MK114856.1": {
+        "qlen": 9477,
         "hypermutation_probablility": 1.0,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9477,
@@ -136,6 +153,7 @@
         "blast_n_conseqs": 4
     },
     "MK115009.1": {
+        "qlen": 9207,
         "hypermutation_probablility": 1.0,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9207,
@@ -144,6 +162,7 @@
         "blast_n_conseqs": 3
     },
     "MK115387.1": {
+        "qlen": 9136,
         "hypermutation_probablility": 0.5412311092694289,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9136,
@@ -152,6 +171,7 @@
         "blast_n_conseqs": 2
     },
     "MK115491.1": {
+        "qlen": 9422,
         "hypermutation_probablility": 0.8951015182445495,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9422,
@@ -160,6 +180,7 @@
         "blast_n_conseqs": 2
     },
     "MK116110.1": {
+        "qlen": 8967,
         "hypermutation_probablility": 0.07021438897893317,
         "inferred_subtype": "Ref.B.TH.90.BK132.AY173951",
         "blast_matched_qlen": 8967,
@@ -168,6 +189,7 @@
         "blast_n_conseqs": 3
     },
     "MK115527.1": {
+        "qlen": 9481,
         "hypermutation_probablility": 0.7689834393883834,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9481,
@@ -176,6 +198,7 @@
         "blast_n_conseqs": 2
     },
     "MK114997.1": {
+        "qlen": 9055,
         "hypermutation_probablility": 0.054959132555391754,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9055,
@@ -184,6 +207,7 @@
         "blast_n_conseqs": 2
     },
     "MK115518.1": {
+        "qlen": 9537,
         "hypermutation_probablility": 0.6385326595592609,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9537,
@@ -192,6 +216,7 @@
         "blast_n_conseqs": 3
     },
     "MK115065.1": {
+        "qlen": 9214,
         "hypermutation_probablility": 0.033517722768753644,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9214,
@@ -200,6 +225,7 @@
         "blast_n_conseqs": 6
     },
     "MK115464.1": {
+        "qlen": 9663,
         "hypermutation_probablility": 1.0,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9663,
@@ -208,6 +234,7 @@
         "blast_n_conseqs": 2
     },
     "MK115530.1": {
+        "qlen": 9544,
         "hypermutation_probablility": 0.5789377103398377,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9544,
@@ -216,6 +243,7 @@
         "blast_n_conseqs": 2
     },
     "MK115520.1": {
+        "qlen": 9589,
         "hypermutation_probablility": 0.5200353682902832,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9589,
@@ -224,6 +252,7 @@
         "blast_n_conseqs": 3
     },
     "MK115503.1": {
+        "qlen": 9617,
         "hypermutation_probablility": 0.4263025132504157,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9617,
@@ -232,6 +261,7 @@
         "blast_n_conseqs": 2
     },
     "MK115570.1": {
+        "qlen": 9485,
         "hypermutation_probablility": 0.738578434638724,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9485,
@@ -240,6 +270,7 @@
         "blast_n_conseqs": 2
     },
     "MK115509.1": {
+        "qlen": 9353,
         "hypermutation_probablility": 0.7866198309713798,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9353,
@@ -248,6 +279,7 @@
         "blast_n_conseqs": 2
     },
     "MK115702.1": {
+        "qlen": 9098,
         "hypermutation_probablility": 0.14401391767451666,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9098,
@@ -256,6 +288,7 @@
         "blast_n_conseqs": 4
     },
     "MK115095.1": {
+        "qlen": 9137,
         "hypermutation_probablility": 1.0,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9137,
@@ -264,6 +297,7 @@
         "blast_n_conseqs": 2
     },
     "MK115490.1": {
+        "qlen": 9347,
         "hypermutation_probablility": 0.8863248655310947,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9347,
@@ -272,6 +306,7 @@
         "blast_n_conseqs": 3
     },
     "MK115576.1": {
+        "qlen": 9266,
         "hypermutation_probablility": 0.818189227062389,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9266,
@@ -280,6 +315,7 @@
         "blast_n_conseqs": 3
     },
     "OQ092466": {
+        "qlen": 9686,
         "hypermutation_probablility": 0.3876036547663967,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9686,
@@ -288,6 +324,7 @@
         "blast_n_conseqs": 3
     },
     "OQ092463": {
+        "qlen": 9605,
         "hypermutation_probablility": 0.21628713708846803,
         "inferred_subtype": "Ref.B.TH.90.BK132.AY173951",
         "blast_matched_qlen": 9605,
@@ -296,6 +333,7 @@
         "blast_n_conseqs": 2
     },
     "OQ092465": {
+        "qlen": 9659,
         "hypermutation_probablility": 0.02412789935966586,
         "inferred_subtype": "Ref.28_BF.BR.99.BREPM12817.DQ085874",
         "blast_matched_qlen": 9659,
@@ -304,6 +342,7 @@
         "blast_n_conseqs": 2
     },
     "OQ092462": {
+        "qlen": 9714,
         "hypermutation_probablility": 0.10777665573070194,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9714,
@@ -312,6 +351,7 @@
         "blast_n_conseqs": 3
     },
     "OQ092464": {
+        "qlen": 9556,
         "hypermutation_probablility": 0.006887768010151674,
         "inferred_subtype": "Ref.28_BF.BR.99.BREPM12817.DQ085874",
         "blast_matched_qlen": 9556,
@@ -320,6 +360,7 @@
         "blast_n_conseqs": 2
     },
     "OQ092467": {
+        "qlen": 9936,
         "hypermutation_probablility": 0.6416537859942263,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9936,
diff --git a/tests/expected-results-single-csv/holistic.csv b/tests/expected-results-single-csv/holistic.csv
index 5567538..f1840a4 100644
--- a/tests/expected-results-single-csv/holistic.csv
+++ b/tests/expected-results-single-csv/holistic.csv
@@ -1,2 +1,2 @@
-seqid,hypermutation_probablility,inferred_subtype,blast_matched_qlen,blast_sseq_coverage,blast_qseq_coverage,blast_n_conseqs
-MN692145,0.1661041079701131,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9689,1.1304660973351168,1.1271545051088863,3
+seqid,qlen,hypermutation_probablility,inferred_subtype,blast_matched_qlen,blast_sseq_coverage,blast_qseq_coverage,blast_n_conseqs
+MN692145,9689,0.1661041079701131,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9689,1.1304660973351168,1.1271545051088863,3
diff --git a/tests/expected-results-single/holistic.json b/tests/expected-results-single/holistic.json
index a491a39..3cd930d 100644
--- a/tests/expected-results-single/holistic.json
+++ b/tests/expected-results-single/holistic.json
@@ -1,5 +1,6 @@
 {
     "MN692145": {
+        "qlen": 9689,
         "hypermutation_probablility": 0.1661041079701131,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9689,
diff --git a/tests/expected-results-small-csv/holistic.csv b/tests/expected-results-small-csv/holistic.csv
index 099a7b3..8da1aa4 100644
--- a/tests/expected-results-small-csv/holistic.csv
+++ b/tests/expected-results-small-csv/holistic.csv
@@ -1,6 +1,6 @@
-seqid,hypermutation_probablility,inferred_subtype,blast_matched_qlen,blast_sseq_coverage,blast_qseq_coverage,blast_n_conseqs
-KX505501.1,0.7087072014754221,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,1997,0.2498199403230785,1.2158237356034052,4
-MN691959,0.19593905853945925,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9493,1.0816956477003807,1.1086063415148004,3
-MN692074,0.36378645339477633,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,4178,0.5041670953801831,1.1728099569171853,4
-MN692145,0.1661041079701131,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9689,1.1304660973351168,1.1271545051088863,3
-MN090335,0.1754017863888554,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9069,0.9842576396748637,1.0603153600176425,3
+seqid,qlen,hypermutation_probablility,inferred_subtype,blast_matched_qlen,blast_sseq_coverage,blast_qseq_coverage,blast_n_conseqs
+KX505501.1,1997,0.7087072014754221,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,1997,0.2498199403230785,1.2158237356034052,4
+MN691959,9493,0.19593905853945925,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9493,1.0816956477003807,1.1086063415148004,3
+MN692074,4178,0.36378645339477633,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,4178,0.5041670953801831,1.1728099569171853,4
+MN692145,9689,0.1661041079701131,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9689,1.1304660973351168,1.1271545051088863,3
+MN090335,9069,0.1754017863888554,Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455,9069,0.9842576396748637,1.0603153600176425,3
diff --git a/tests/expected-results-small/holistic.json b/tests/expected-results-small/holistic.json
index ba48ccb..2bbd7d0 100644
--- a/tests/expected-results-small/holistic.json
+++ b/tests/expected-results-small/holistic.json
@@ -1,5 +1,6 @@
 {
     "KX505501.1": {
+        "qlen": 1997,
         "hypermutation_probablility": 0.7087072014754221,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 1997,
@@ -8,6 +9,7 @@
         "blast_n_conseqs": 4
     },
     "MN691959": {
+        "qlen": 9493,
         "hypermutation_probablility": 0.19593905853945925,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9493,
@@ -16,6 +18,7 @@
         "blast_n_conseqs": 3
     },
     "MN692074": {
+        "qlen": 4178,
         "hypermutation_probablility": 0.36378645339477633,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 4178,
@@ -24,6 +27,7 @@
         "blast_n_conseqs": 4
     },
     "MN692145": {
+        "qlen": 9689,
         "hypermutation_probablility": 0.1661041079701131,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9689,
@@ -32,6 +36,7 @@
         "blast_n_conseqs": 3
     },
     "MN090335": {
+        "qlen": 9069,
         "hypermutation_probablility": 0.1754017863888554,
         "inferred_subtype": "Ref.B.FR.83.HXB2_LAI_IIIB_BRU.K03455",
         "blast_matched_qlen": 9069,