Close #806, by adding extra references for HIVGHA project.

cfe-lab · Feb 7, 2022 · 3c3f0f2 · 3c3f0f2
1 parent 3ceb954
commit 3c3f0f2
Show file tree

Hide file tree

Showing 7 changed files with 578 additions and 36 deletions.
diff --git a/micall/blast_db/make_blast_db.py b/micall/blast_db/make_blast_db.py
@@ -29,6 +29,9 @@ def make_blast_db(projects_json, refs_fasta):
         for name, region in projects['regions'].items():
             if region['seed_group'] is None:
                 continue
+            if region['seed_group'] == 'HIVGHA-seed':
+                # Exclude this project, because they're recombinant.
+                continue
             if name == 'HIV1-CON-XX-Consensus-seed':
                 # Only used by G2P alignment.
                 continue

diff --git a/micall/drivers/sample.py b/micall/drivers/sample.py
@@ -12,6 +12,7 @@
 from micall.core.coverage_plots import coverage_plot
 from micall.core.plot_contigs import plot_genome_coverage
 from micall.core.prelim_map import prelim_map
+from micall.core.project_config import ProjectConfig
 from micall.core.remap import remap, map_to_contigs
 from micall.core.sam2aln import sam2aln
 from micall.core.trim_fastqs import trim
@@ -22,6 +23,16 @@
 logger = logging.getLogger(__name__)
 
 
+def exclude_extra_seeds(excluded_seeds: typing.Sequence[str],
+                        project_code: str = None) -> typing.Sequence[str]:
+    if project_code == 'HIVGHA':
+        return excluded_seeds
+    projects = ProjectConfig.loadDefault()
+    hivgha_seeds = projects.getProjectSeeds('HIVGHA')
+    hiv_seeds = projects.getProjectSeeds('HIV')
+    return sorted((hivgha_seeds - hiv_seeds) | set(excluded_seeds))
+
+
 class Sample:
     def __init__(self,
                  basespace_id=None,
@@ -126,6 +137,8 @@ def process(self,
         use_gzip = force_gzip or self.fastq1.endswith('.gz')
 
         sample_info = self.load_sample_info()
+        excluded_seeds = exclude_extra_seeds(excluded_seeds,
+                                             sample_info.get('project'))
 
         with open(self.read_summary_csv, 'w') as read_summary:
             trim((self.fastq1, self.fastq2),

diff --git a/micall/project_scoring.json b/micall/project_scoring.json
@@ -6106,7 +6106,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -6237,7 +6239,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -6359,7 +6363,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -6481,7 +6487,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -6603,7 +6611,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -6725,7 +6735,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -6847,7 +6859,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -6969,7 +6983,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -7168,7 +7184,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -7347,7 +7365,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -7582,7 +7602,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         },
         {
@@ -7704,7 +7726,9 @@
             "HIV1-O-SN-AJ302646-seed",
             "HIV1-O-US-JN571034-seed",
             "HIV1-P-CM-HQ179987-seed",
-            "HIV1-P-FR-GU111555-seed"
+            "HIV1-P-FR-GU111555-seed",
+            "HIV1-CRF02_AG-GH-AB286855-seed",
+            "HIV1-CRF06_CPX-GH-AB286851-seed"
           ]
         }
       ]