From f65cada4f70d7449f38f1fef5164d7c52283a0e9 Mon Sep 17 00:00:00 2001
From: Pjotr Prins
Date: Sun, 19 Nov 2023 17:14:15 +0100
Subject: Iterating through hits

---
 gn/data/dataset.scm                    | 23 +++++++++++++++++++++++
 gn/data/hits.scm                       |  2 +-
 scripts/precompute/precompute-hits.scm | 22 ++++++++++++++++++----
 3 files changed, 42 insertions(+), 5 deletions(-)
 create mode 100644 gn/data/dataset.scm

diff --git a/gn/data/dataset.scm b/gn/data/dataset.scm
new file mode 100644
index 0000000..5e81d1e
--- /dev/null
+++ b/gn/data/dataset.scm
@@ -0,0 +1,23 @@
+(define-module (gn data dataset)
+  #:use-module (json)
+  #:use-module (ice-9 match)
+  #:use-module (ice-9 format)
+  #:use-module (ice-9 iconv)
+  #:use-module (ice-9 receive)
+  #:use-module (ice-9 string-fun)
+  ;; #:use-module (gn db sparql)
+  #:use-module (dbi dbi)
+  #:use-module (gn db mysql)
+  #:use-module (gn data group)
+  #:use-module (gn util convert)
+  #:use-module (web gn-uri)
+
+  #:export (
+            dataset-name
+            ))
+
+(define (dataset-name db probesetfreeze-id)
+  (dbi-query db
+             (string-append
+              "select Name2 from ProbeSetFreeze where Id=" (int-to-string probesetfreeze-id) " limit 1;"))
+  (assoc-ref (get-row db) "Name2"))
diff --git a/gn/data/hits.scm b/gn/data/hits.scm
index a66eee5..f0a24f3 100644
--- a/gn/data/hits.scm
+++ b/gn/data/hits.scm
@@ -17,5 +17,5 @@
             ))
 
 (define (get-next-hit-for-precompute db prev-id)
-  (dbi-query db (string-append "select Locus, DataId, ProbeSetId from ProbeSetXRef where DataId>" (int-to-string prev-id) " AND Locus_old is NULL ORDER BY DataId LIMIT 1"))
+  (dbi-query db (string-append "select Locus, DataId, ProbeSetId, ProbeSetFreezeId from ProbeSetXRef where DataId>" (int-to-string prev-id) " AND Locus_old is NULL ORDER BY DataId LIMIT 1"))
   (get-row db))
diff --git a/scripts/precompute/precompute-hits.scm b/scripts/precompute/precompute-hits.scm
index 0088431..d8170db 100755
--- a/scripts/precompute/precompute-hits.scm
+++ b/scripts/precompute/precompute-hits.scm
@@ -6,6 +6,7 @@
 
 (use-modules (dbi dbi)
              (gn db mysql)
+             (gn data dataset)
              (gn data hits)
              (gn data strains)
              (gn util convert)
@@ -35,7 +36,7 @@
 
      ;(newline)
      (define bxd-strains (bxd-strain-id-names #:map? #t))
-     ;(display (assoc 64728 bxd-strains))
+                                        ;(display (assoc 64728 bxd-strains))
      ;(newline)
      ;(newline)
      ;(dbi-query db "SELECT * FROM ProbeSetXRef LIMIT 3")
@@ -47,15 +48,26 @@
      ;; ---- get first available dataset for precompute:
      ;; @@ order by dataid - recurse
 
+     (define (get-trait-name db probeset-id)
+       (dbi-query db (string-append "select Name,Symbol from ProbeSet where Id=" (int-to-string probeset-id) " limit 1"))
+       (get-row db)
+   )
 
      (define (run-precompute db prev-id)
        (let* [(hit (get-next-hit-for-precompute db prev-id))
-              (data-id (assoc-ref hit "DataId"))]
+              (data-id (assoc-ref hit "DataId"))
+              (data-id-str (int-to-string data-id))
+              (probesetfreeze-id (assoc-ref hit "ProbeSetFreezeId"))
+              (probeset-id (assoc-ref hit "ProbeSetId"))
+              (trait (get-trait-name db probeset-id))
+              (trait-name (assoc-ref trait "Name"))
+              ]
          (display hit)
          (display data-id)
          (newline)
+         (define name (dataset-name db probesetfreeze-id))
          ;; ---- Get strains and phenotypes for this dataset
-         (dbi-query db (string-append "SELECT StrainId,value from ProbeSetData WHERE Id=" (int-to-string data-id)))
+         (dbi-query db (string-append "SELECT StrainId,value from ProbeSetData WHERE Id=" data-id-str))
          (define traits (get-rows-apply db
                                         (lambda (r) `(,(assoc-ref r "StrainId") . ,(assoc-ref r "value")))
                                         '()))
@@ -71,10 +83,12 @@
                           traits))
          (if (= 0 (length non-bxd))
              (begin
-               (display "WE HAVE OUR BXD DATASET for precompute!\n")
+               (if name
+                   (display (string-append "WE HAVE OUR BXD DATASET " name " and trait " trait-name " for precompute!\n")))
                (display data-id)
                (display traits)
                (newline)
+
                ))
          (run-precompute db data-id)
          ))
-- 
cgit v1.2.3