Merge pull request #154 from XingerTang/feat_recomb_new

Provide map file input option for non-hybrid mode #153

docs/source/usage.rst

@@ -25,6 +25,7 @@ Input Arguments
       -start_snp START_SNP
                           The first marker to consider. The first marker is "1". Default: 1.
       -stop_snp STOP_SNP  The last marker to consider. Default: all markers considered.
+      -map_file MAP_FILE  A map file for all loci.
 
 |Software| requires a pedigree file (``-ped_file``) and one or more genomic data files to run the analysis.
 
@@ -137,8 +138,7 @@ Hybrid peeling arguments
     Single locus arguments:
       -seg_file SEG_FILE    A segregation probabilities file for hybrid peeling.
       -seg_map_file SEG_MAP_FILE
-                            A map file for loci in the segregation probabilities file.
-      -map_file MAP_FILE    A map file for all loci in hybrid peeling.
+                            A map file for loci in the segregation probabilities file in hybrid peeling.
 
 In order to run hybrid peeling the user needs to supply a ``-map_file`` which gives the genetic positions for the SNPs in the sequence allele read counts data supplied, a ``-seg_map_file`` which gives the genetic position for the SNPs in the segregation file, and a ``-seg_file`` which gives the segregation values generated via multi-locus iterative peeling. These arguments are not required for running in multi-locus mode.
 

src/tinypeel/Peeling/PeelingInfo.py

@@ -5,6 +5,7 @@
 
 from ..tinyhouse import ProbMath
 from ..tinyhouse import HaplotypeOperations
+from ..tinyhouse import InputOutput
 
 
 #####################################################################
@@ -22,8 +23,11 @@ def createPeelingInfo(pedigree, args, createSeg=True, phaseFounder=False):
         nInd=pedigree.maxIdn, nFam=pedigree.maxFam, nLoci=nLoci, createSeg=createSeg
     )
     peelingInfo.isSexChrom = args.sex_chrom
-    # Information about the peeling positions are handled elsewhere.
-    peelingInfo.positions = None
+    if args.map_file:
+        peelingInfo.positions = np.array(
+            InputOutput.readMapFile(args.map_file, args.startsnp, args.stopsnp)[2],
+            dtype=np.int64,
+        )
 
     # Generate the segregation tensors.
     peelingInfo.segregationTensor = ProbMath.generateSegregation(e=1e-06)
@@ -102,9 +106,9 @@ def setupTransmission(length, peelingInfo):
     if peelingInfo.positions is None:
         localMap = np.linspace(0, 1, num=peelingInfo.nLoci, dtype=np.float32)
     else:
-        localMap = (
-            peelingInfo.positions / peelingInfo.positions[-1]
-        )  # This should be sorted. Need to add in code to check.
+        localMap = (peelingInfo.positions - peelingInfo.positions[0]) / (
+            peelingInfo.positions[-1] - peelingInfo.positions[0]
+        )
     for i in range(peelingInfo.nLoci - 1):
         distance = localMap[i + 1] - localMap[i]
         distance = distance * length
@@ -213,7 +217,7 @@ def getHetMidpoint(geno):
 spec["transmissionRate"] = optional(float32[:])
 spec["maf"] = optional(float32[:])
 
-spec["positions"] = optional(float32[:])  # Not sure we use this.
+spec["positions"] = optional(int64[:])
 spec["iteration"] = int64
 
 

src/tinypeel/tinypeel.py

@@ -189,9 +189,7 @@ def getLociAndDistance(snpMap, segMap):
 def generateSingleLocusSegregation(peelingInfo, pedigree, args):
     if args.segfile is not None:
         # This just gets the locations in the map files.
-        snpMap = np.array(
-            InputOutput.readMapFile(args.map_file, args.startsnp, args.stopsnp)[2]
-        )
+        snpMap = peelingInfo.positions
         segMap = np.array(InputOutput.readMapFile(args.seg_map_file)[2])
 
         loci, distance = getLociAndDistance(snpMap, segMap)
@@ -387,6 +385,13 @@ def getArgs():
             "stopsnp",
         ],
     )
+    input_parser.add_argument(
+        "-map_file",
+        default=None,
+        required=False,
+        type=str,
+        help="A map file for all loci.",
+    )
 
     # Output options
     output_parser = parser.add_argument_group("Output Options")
@@ -534,19 +539,12 @@ def getArgs():
     )
 
     singleLocus_parser = parser.add_argument_group("Hybrid peeling arguments")
-    singleLocus_parser.add_argument(
-        "-map_file",
-        default=None,
-        required=False,
-        type=str,
-        help="a map file for all loci in hybrid peeling.",
-    )
     singleLocus_parser.add_argument(
         "-seg_map_file",
         default=None,
         required=False,
         type=str,
-        help="a map file for loci in the segregation probabilities file.",
+        help="A map file for loci in the segregation probabilities file in hybrid peeling.",
     )
     singleLocus_parser.add_argument(
         "-seg_file",

tests/functional_tests/run_func_test.py

@@ -68,7 +68,6 @@ def delete_columns(two_d_list, col_del):
 
 class TestClass:
     path = os.path.join("tests", "functional_tests")
-    command = "AlphaPeel "
     test_cases = None
     input_file_depend_on_test_cases = None
 
@@ -96,6 +95,8 @@ def generate_command(self):
         """
         generate the command for the test
         """
+        self.command = "AlphaPeel "
+
         for file in self.input_files:
             if (
                 (self.test_cases is not None)
@@ -539,6 +540,47 @@ def test_out_id_only(self):
         for ind in self.output:
             assert "MotherOf" not in ind[0] and "FatherOf" not in ind[0]
 
+    def test_map_input(self):
+        """
+        Run the test for the input map file
+        """
+        self.test_name = "test_map_input"
+        self.prepare_path()
+
+        self.arguments = {"method": "multi"}
+        self.output_file_to_check = "dosage"
+
+        # without map file input
+        self.input_files = ["geno_file", "ped_file"]
+        self.output_file_prefix = "map_input.no_map_file"
+
+        self.generate_command()
+        os.system(self.command)
+
+        self.output_file_path = os.path.join(
+            self.output_path,
+            f"{self.output_file_prefix}.{self.output_file_to_check}.txt",
+        )
+
+        self.first_output = read_and_sort_file(self.output_file_path)
+
+        # with map file input
+        self.input_files.append("map_file")
+        self.output_file_prefix = "map_input.with_map_file"
+
+        self.generate_command()
+        os.system(self.command)
+
+        self.output_file_path = os.path.join(
+            self.output_path,
+            f"{self.output_file_prefix}.{self.output_file_to_check}.txt",
+        )
+
+        self.second_output = read_and_sort_file(self.output_file_path)
+
+        # the two outputs should match
+        assert self.first_output == self.second_output
+
     # TODO test_plink for PLINK
     #      a. binary PLINK output
     #      b. binary output + input

tests/functional_tests/test_map_input/geno_file.txt

@@ -0,0 +1,6 @@
+M0 1 2 1 0 9
+F0 1 2 0 1 2
+M1 0 2 0 1 2
+F1 2 2 1 0 2
+M2 1 2 1 1 2
+F2 1 2 0 0 2

tests/functional_tests/test_map_input/map_file.txt

@@ -0,0 +1,5 @@
+1 1-1 1
+1 1-2 2
+1 1-3 3
+1 1-4 4
+1 1-5 5

tests/functional_tests/test_map_input/ped_file.txt

@@ -0,0 +1,6 @@
+M0 0 0
+F0 0 0
+M1 M0 F0
+F1 M0 F0
+M2 M1 F1
+F2 M1 F1