From f2e8c637307146a6796c56b0d654427f9c567aa8 Mon Sep 17 00:00:00 2001
From: Pjotr Prins
Date: Sun, 7 Jul 2024 15:17:48 +0200
Subject: precompute: making sure order is correct - validating results

---
 scripts/precompute/list-traits-to-compute.scm | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)

(limited to 'scripts/precompute/list-traits-to-compute.scm')

diff --git a/scripts/precompute/list-traits-to-compute.scm b/scripts/precompute/list-traits-to-compute.scm
index 61ce755..fbe89a6 100755
--- a/scripts/precompute/list-traits-to-compute.scm
+++ b/scripts/precompute/list-traits-to-compute.scm
@@ -131,7 +131,8 @@ The following is produced by gemma-wrapper as metadata
                                (time . ,(date->string (time-utc->date (current-time))))))
                       (data .
                             ((,id .
-                                  ((probesetfreeze-id . ,probesetfreeze-id)
+                                  ((group . "BXD")
+                                   (probesetfreeze-id . ,probesetfreeze-id)
                                    (probeset-id . ,probeset-id)
                                    (name . ,name)
                                    (trait-name . ,trait-name)
@@ -150,6 +151,7 @@ The following is produced by gemma-wrapper as metadata
      ;; (let [(bxd-strains (memo-bxd-strain-id-names #:used-for-mapping? #t))]
        (display "writing-phenotypes...")
        (define (run-list-traits-to-compute db num prev-id)
+         ;; ---- Build a query to collect num traits
          (let* [(count (if (< batch-size num)
                            batch-size
                            num))
@@ -164,7 +166,7 @@ The following is produced by gemma-wrapper as metadata
                 ]
            (display query)
            (dbi-query db query)
-           ;; ---- Walk each resulting trait
+           ;; ---- Walk each resulting trait and build a hash of data-id and list of trait values
            (let [(id-traits (get-rows db '()))
                  (nrecs '())]
              (for-each (lambda (r)
@@ -178,21 +180,20 @@ The following is produced by gemma-wrapper as metadata
                                 )]
                       (set! nrecs (assoc-set! nrecs data-id lst))))
                        id-traits)
-             ;; --- create the json output as a file
-             (for-each (lambda (r h)
+             ;; --- create the json output as a file by walking traits and hits
+             (for-each (lambda (h)
                          (let* [
+                                (id (hit-data-id h))
                                 (probeset-id (hit-probeset-id h))
-                                ;; (data-id-str (int-to-string id))
                                 (probesetfreeze-id (hit-probesetfreeze-id h))
                                 (name (dataset-name db probesetfreeze-id))
                                 (trait (get-trait db probeset-id))
                                 (trait-name (assoc-ref trait "Name"))
+                                (recs (assoc-ref nrecs id))
                                 ]
-                           (match r
-                             ((id . recs) (if (has-bxd? recs)
+                             (if (has-bxd? recs)
                                               (write-json-ld id name trait trait-name probesetfreeze-id probeset-id recs)
-                                              ))
-                             ))) nrecs hits)
+                             ))) hits)
              (if (> rest 0)
                  (run-list-traits-to-compute db rest (first (reverse data-ids)))) ;; start precompute
            )))
-- 
cgit v1.2.3