Add second integration test and fix bugs discovered

Author: reedacartwright

sofos.hpp

@@ -65,11 +65,10 @@ class SofosHistogram {
     void ZeroCounts();
 
     size_t num_rows() const {
-        assert(prior_.size() == size_ + 1);
-        assert(observed_.size() == size_ + 1);
-        assert(posterior_.size() == size_ + 1);
+        assert(prior_.size() == posterior_.size());
+        assert(observed_.size() == posterior_.size());
 
-        return size_ + 1;
+        return posterior_.size();
     }
 
     std::array<double, 3> row(size_t i) const {

test/run_tests.bash

@@ -1,5 +1,7 @@
 #!/bin/bash
 
-SOFOS=$1
+SOFOS="${1:-../sofos}"
 
-bash test-01.bash "${SOFOS}"
+bash test-01.bash "${SOFOS}" || exit 1
+
+bash test-02.bash "${SOFOS}" || exit 1

test/test-02.bash

@@ -0,0 +1,93 @@
+#!/bin/bash
+#
+# library(rmutil)
+# f = function(n,a,b) dbetabinom(0:n,n,a/(a+b),a+b)
+
+echo "Running Integration Test: folded+refalt ..."
+
+SOFOS="${1:-../sofos}"
+VCF=test-default.vcf
+
+output=$(${SOFOS} -q -a 0.1 -b 1 -n 10 -f -r ${VCF})
+
+hnum=$(echo "${output}" | grep '^#' | wc -l)
+if [[ 10 -ne "${hnum}" ]]; then
+	echo "  ERROR: Heading has ${hnum} lines instead of 10."
+	exit 1
+fi
+header=$(echo "${output}" | awk 'NR==11')
+if [[ $header != "Number,Prior,Observed,Posterior" ]]; then
+	echo "  ERROR: Header has wrong column names."
+	exit 1
+fi
+
+nrow=$(echo "${output}" | grep -v '^#' | awk -F, 'END {print NR}')
+if [[ 7 -ne "${nrow}" ]]; then
+	echo "  ERROR: Body has wrong number of rows."
+	exit 1
+fi
+
+# Check posterior
+expected="0.613042662838015
+0.481731622579739
+0.354864195501706
+0.259140144535515
+0.200677862510758
+0.0905435120342663"
+
+echo "${output}" | grep -v '^#' | awk -F, 'NR>1 {print $4}' \
+	| paste -d, - <(echo "${expected}") \
+	| awk -F, -v m=1e-7 'function abs(v) {return v < 0 ? -v : v} {a=abs(($1-$2))/$2; if(a > m){m = a} } END{exit (m > 1e-7) }' 
+if [[ $? -ne 0 ]]; then
+	echo "  ERROR: Maximum relative error in posterior exceeds 1e-7."
+	exit 1
+fi
+
+# Check prior
+expected="1.52304554473485
+0.172084774253705
+0.106856638049549
+0.0851179668121595
+0.076115591667104
+0.0367794844826313"
+
+echo "${output}" | grep -v '^#' | awk -F, 'NR>1 {print $2}' \
+	| paste -d, - <(echo "${expected}") \
+	| awk -F, -v m=1e-7 'function abs(v) {return v < 0 ? -v : v} {a=abs(($1-$2))/$2; if(a > m){m = a} } END{exit (m > 1e-7) }' 
+if [[ $? -ne 0 ]]; then
+	echo "  ERROR: Maximum relative error in prior exceeds 1e-7."
+	exit 1
+fi
+
+# Check obs
+expected="0
+0
+2
+0
+0
+0
+"
+
+echo "${output}" | grep -v '^#' | awk -F, 'NR>1 {print $3}' \
+	| paste -d, - <(echo "${expected}") \
+	| awk -F, -v m=1e-7 'function abs(v) {return v < 0 ? -v : v} {a=abs(($1-$2)); if(a > m){m = a} } END{exit (m > 1e-7) }' 
+if [[ $? -ne 0 ]]; then
+	echo "  ERROR: Maximum error in obs exceeds 1e-7."
+	exit 1
+fi
+
+# Check indexes
+expected="0
+1
+2
+3
+4
+5"
+
+echo "${output}" | grep -v '^#' | awk -F, 'NR>1 {print $1}' \
+	| paste -d, - <(echo "${expected}") \
+	| awk -F, -v m=1e-7 'function abs(v) {return v < 0 ? -v : v} {a=abs(($1-$2)); if(a > m){m = a} } END{exit (m > 1e-7) }' 
+if [[ $? -ne 0 ]]; then
+	echo "  ERROR: Maximum error in obs exceeds 1e-7."
+	exit 1
+fi

unittest.cc

@@ -790,30 +790,52 @@ TEST_CASE("output_header generates a header containing program information") {
 TEST_CASE("output_body generates a csv table containing column names and values") {
     sofos_params_t params;
     params.size = 4;
-    Sofos sofos{params};
-    std::stringstream str;
-    output_body(str, sofos);
 
-    std::vector<std::vector<std::string>> table;
-    std::string line;
-    while(std::getline(str, line, '\n')) {
-        std::stringstream row;
-        row.str(line);
-        std::string token;
-        std::vector<std::string> tokens;
-        while(std::getline(row, token, ',')) {
-            tokens.push_back(token);
+    std::stringstream str;
+    auto parse_table = [&]() -> std::vector<std::vector<std::string>> {
+        std::vector<std::vector<std::string>> table;
+        std::string line;
+        while(std::getline(str, line, '\n')) {
+            std::stringstream row;
+            row.str(line);
+            std::string token;
+            std::vector<std::string> tokens;
+            while(std::getline(row, token, ',')) {
+                tokens.push_back(token);
+            }
+            table.push_back(tokens);
         }
-        table.push_back(tokens);
-    }
+        return table;        
+    };
 
-    REQUIRE(table.size() == 6);
-    CHECK(table[0].size() == 4);
-    CHECK(table[1].size() == 4);
-    CHECK(table[2].size() == 4);
-    CHECK(table[3].size() == 4);
-    CHECK(table[4].size() == 4);
-    CHECK(table[5].size() == 4);
+    SECTION("when histogram is unfolded"){
+        params.flag_folded = false;       
+        Sofos sofos{params};
+        sofos.FinishHistogram();
+        output_body(str, sofos);
+        auto table = parse_table();
+
+        REQUIRE(table.size() == 6);
+        CHECK(table[0].size() == 4);
+        CHECK(table[1].size() == 4);
+        CHECK(table[2].size() == 4);
+        CHECK(table[3].size() == 4);
+        CHECK(table[4].size() == 4);
+        CHECK(table[5].size() == 4);
+    }
+    SECTION("when histogram is folded"){
+        params.flag_folded = true;
+        Sofos sofos{params};
+        sofos.FinishHistogram();
+        output_body(str, sofos);
+        auto table = parse_table();
+
+        REQUIRE(table.size() == 4);
+        CHECK(table[0].size() == 4);
+        CHECK(table[1].size() == 4);
+        CHECK(table[2].size() == 4);
+        CHECK(table[3].size() == 4);
+    }
 }
 
 TEST_CASE("Retrieving values from a VCF file") {