Merge pull request #15 from cokelaer/main

automatically set the index algorithm

README.rst

@@ -130,6 +130,8 @@ Changelog
 ========= ======================================================================
 Version   Description
 ========= ======================================================================
+1.0.1     * automatically fill the bwa index algorithm and fix bwa_index rule to 
+            use the options in the config file (not the harcoded one)
 1.0.0     * use last warppers and graphviz apptainer
 0.12.0    * set all apptainers containers and add vcf to bcf conversions
           * Update rule sambamba to use latest wrappers

sequana_pipelines/variant_calling/main.py

@@ -63,11 +63,11 @@ def __init__(self, prog=NAME, epilog=None):
             help="The annotation for snpeff"
         )
         pipeline_group.add_argument("--threads", dest="threads", default=4, type=int)
-        pipeline_group.add_argument("--do-coverage", dest="do_coverage", 
-            action="store_true", 
+        pipeline_group.add_argument("--do-coverage", dest="do_coverage",
+            action="store_true",
             help="perform the coverage analysis using sequana_coverage")
-        pipeline_group.add_argument("--do-joint-calling", dest="do_joint_calling", 
-            action="store_true", 
+        pipeline_group.add_argument("--do-joint-calling", dest="do_joint_calling",
+            action="store_true",
             help="do the joint calling analysise")
 
         pipeline_group.add_argument("-o", "--circular", action="store_true")
@@ -136,12 +136,24 @@ def main(args=None):
         cfg['sequana_coverage']["circular"] = options.circular
         cfg['joint_freebayes']['do'] = options.do_joint_calling
         cfg['bwa_mem']['threads'] = options.threads
+
+
         cfg['freebayes']['ploidy'] = options.freebayes_ploidy
 
         cfg.reference_file = os.path.abspath(options.reference)
         manager.exists(cfg.reference_file)
 
-        # coverage
+        # Given the reference, let us compute its length and st the index algorithm
+        from sequana import FastA
+        f = FastA(options.reference)
+        N = f.get_stats()['total_length']
+
+        # seems to be a hardcoded values in bwa according to the documentation
+        if N >= 2000000000:
+            cfg['bwa_mem']['index_algorithm'] = "bwtsw"
+        else:
+            cfg['bwa_mem']['index_algorithm'] = "is"
+
 
 
     # finalise the command and save it; copy the snakemake. update the config

sequana_pipelines/variant_calling/variant_calling.rules

@@ -116,8 +116,8 @@ if not config["input_pattern"].endswith("bam"):
         log:
             "reference/build.log"
         params:
-            options="",
-            index_algorithm="is"
+            options=config['bwa_mem'].get('index_options', ""),
+            index_algorithm=config['bwa_mem'].get('index_algorithm', "is")
         container:
             config['apptainers']['sequana_tools']
         threads: 2

setup.py

@@ -6,7 +6,7 @@
 
 _MAJOR               = 1
 _MINOR               = 0
-_MICRO               = 0
+_MICRO               = 1
 
 version              = '%d.%d.%d' % (_MAJOR, _MINOR, _MICRO)
 release              = '%d.%d' % (_MAJOR, _MINOR)