gistfile1.clj

;; actually makes the call to c/first-n to create the limited sorted
;; sample

(defn top-n-seq [sample-name n]
  (let [fields ["?aa" "?v" "?j" "?name" "?nc"]
        sample-name-seqs (<- fields
                             (tcr-seqs :>> fields)
                             (= ?name sample-name))]
    (c/first-n sample-name-seqs n :sort ["?nc"] :reverse true))) 

;; makes two sample sets from tcr-seqs and finds matches 

(defn matching-seqs-limited [x-name y-name x-count y-count]
  (let [seqs-x (top-n-seqs x-name x-count)
        seqs-y (top-n-seqs y-name y-count)]
    (<- [?aa ?v ?j ?x-name ?y-name]
        (seqs-x ?aa ?v ?j ?x-name _)
        (seqs-y ?aa ?v ?j ?y-name _))))

;; a regular clojure function that makes a query, 
;; executes it, and returns a tuple (vector) result
;;
;; counts the number of matches returned by 'matches'
;;
;; I added 'fake-generator' so I can get a 0 for ?total-count 
;; when 'matches' is empty.

(defn count-matching-seqs-limited [x-name y-name x-count y-count]
  (let [matches (matching-seqs-limited x-name y-name x-count y-count)
        fake-generator [[x-name y-name]]
        query-res (??<- [?x-name ?y-name ?total-count]
                      (fake-generator ?x-name ?y-name)
                      (matches !!aa _ _ ?x-name ?y-name)
                      (c/!count !!aa :> ?total-count))]
    (first query-res)))

;; 'source-query' gives me tuples with 2 names and 2 counts.
;; The counts are for how many tuples to sample from 'tcr-seqs'
;; with the respective sample name

(def names-with-match-count
    (<- [?x-name ?y-name ?match-count]
        (source-query ?x-name ?y-name ?x-count ?y-count)
        (count-matching-seqs-limited 
          :< ?x-name ?y-name ?x-count ?y-count 
          :> _ _ ?match-count)))