From 131e0ad6e0279238b747dc1b16070da2109d42eb Mon Sep 17 00:00:00 2001
From: Niyarin <inoanet@gmail.com>
Date: Wed, 7 Jun 2023 10:26:35 +0900
Subject: [PATCH] Add validator for writing vcf.

---
 src/cljam/io/vcf/util/validator.clj       | 62 +++++++++++++++++++++
 test/cljam/io/vcf/util/validator_test.clj | 66 +++++++++++++++++++++++
 2 files changed, 128 insertions(+)
 create mode 100644 src/cljam/io/vcf/util/validator.clj
 create mode 100644 test/cljam/io/vcf/util/validator_test.clj

diff --git a/src/cljam/io/vcf/util/validator.clj b/src/cljam/io/vcf/util/validator.clj
new file mode 100644
index 00000000..8ca33cd5
--- /dev/null
+++ b/src/cljam/io/vcf/util/validator.clj
@@ -0,0 +1,62 @@
+(ns cljam.io.vcf.util.validator
+  (:require [cljam.io.vcf.util :as vcf-util]))
+
+(defn- valid-ref? [s]
+  (or (nil? s)
+      (every? #{\A \T \G \C \N
+                \a \t \g \c \n} s)))
+
+(defn- valid-qual? [v]
+  (or (nil? v)
+      (float? v)
+      (integer? v)))
+
+(defn- check-base-records [variant valid-contig? valid-format?]
+  (cond-> nil
+    (not (valid-contig? (:chr variant))) (assoc :chr "Invalid chr")
+    (not (integer? (:pos variant))) (assoc :pos "Invalid pos")
+    (not (valid-ref? (:ref variant))) (assoc :ref "Invalid ref")
+    (not (every? #(vcf-util/inspect-allele % (:ref variant)) (:alt variant)))
+    (assoc :alt "Invalid alt")
+
+    (not (valid-qual? (:qual variant))) (assoc :qual "Invalid ref")
+    (not (valid-format? (:format variant)))
+    (assoc :format ["Invalid format" (:foramt variant) valid-format?])))
+
+(defn- check-entry-type [entry type-str]
+  ((case type-str
+     "Integer" integer?
+     "Float" #(or (float? %) (integer? %))
+     "Character" char?
+     "String" string?
+     "Flag" (constantly true)
+     (constantly false))
+   entry))
+
+(defn- check-each-samples [variant samples vformat]
+  (reduce
+   (fn [res [ks v]] (assoc-in res ks v))
+   nil
+   (for [sample samples
+         [fmt {:keys [type number]}] vformat
+         :let [entry (get-in variant [sample fmt])
+               entries (if (= number 1) [entry] entry)
+               type-check (and entry
+                               (not (every? #(check-entry-type
+                                              % type) entry)))
+               number-check (and entry (not= (count entries) number))]
+         :when (or type-check number-check)]
+     [[sample fmt]
+      (cond
+        type-check (format
+                    "Not match type declaration. Requires %s , but %s."
+                    type (str entries))
+        number-check (format
+                      "Invalid number of elements. Requires %d , but %d."
+                      number (count entries)))])))
+
+(defn invalid-variant? [variant samples {:keys [contig format]}]
+  (let [valid-contig? #(or (nil? %) ((set (map :id contig)) %))
+        valid-format? #(or (nil? %) ((set (map (comp keyword :id) format)) %))]
+    (merge (check-base-records variant valid-contig? valid-format?)
+           (check-each-samples variant samples format))))
diff --git a/test/cljam/io/vcf/util/validator_test.clj b/test/cljam/io/vcf/util/validator_test.clj
new file mode 100644
index 00000000..401f9abd
--- /dev/null
+++ b/test/cljam/io/vcf/util/validator_test.clj
@@ -0,0 +1,66 @@
+(ns cljam.io.vcf.util.validator-test
+  (:require
+   [clojure.test :refer [deftest is testing]]
+   [cljam.io.vcf.util.validator :as validator]))
+
+(deftest check-each-samples-test
+  (testing "type check"
+    (is (not
+         (#'validator/check-each-samples
+          {:normal {:DP 100}} [:normal]
+          {:DP {:type "Integer" :number 1}})))
+
+    (is (string?
+         (get-in (#'validator/check-each-samples
+                  {:normal {:DP "100"}} [:normal]
+                  {:DP {:type "Integer" :number 1}})
+                 [:normal :DP])))
+
+    (is (not
+         (#'validator/check-each-samples
+          {:normal {:GT "1|0"}} [:normal]
+          {:GT {:type "String" :number 1}})))
+    (is (not
+         (#'validator/check-each-samples
+          {:normal {:A [1 2]}} [:normal]
+          {:A {:type "Integer" :number 2}}))))
+
+  (testing "number check"
+    (is (not
+         (#'validator/check-each-samples
+          {:normal {:A [1 2]}} [:normal]
+          {:A {:type "Integer" :number 2}})))
+
+    (is (string?
+         (get-in
+          (#'validator/check-each-samples
+           {:normal {:A [1]}} [:normal]
+           {:A {:type "Integer" :number 2}})
+          [:normal :A])))))
+
+(deftest invalid-variant?-test
+  (testing "contig check"
+    (is (not
+         (validator/invalid-variant?
+          {:chr "chr10" :pos 10}
+          nil
+          {:contig [{:id "chr10"}]}))))
+  (testing "alt check"
+    (is (not
+         (validator/invalid-variant?
+          {:chr "chr10" :pos 10 :alt ["A" "<DEL>" nil]}
+          nil
+          {:contig [{:id "chr10"}]})))
+    (is (not
+         (validator/invalid-variant?
+          {:chr "chr10" :pos 10 :alt ["G]17:198982]"
+                                      "]13:123456]T"]}
+          nil
+          {:contig [{:id "chr10"}]}))))
+  (testing "qual check"
+    (is (not
+         (validator/invalid-variant?
+          {:chr "chr10" :pos 10
+           :qual 1}
+          nil
+          {:contig [{:id "chr10"}]})))))