From b52e3ef75e5f5b79cba1bacc642f0a13ad33e072 Mon Sep 17 00:00:00 2001
From: Robert Bossy <Robert.Bossy@inra.fr>
Date: Tue, 12 Apr 2022 14:15:07 +0200
Subject: [PATCH] separated taxonomy rewrite and trie compilation

---
 README.md                  | 20 ++++++++++++++++++--
 compile-trie.snakefile     | 19 +++++++++++++++++++
 rewrite-taxonomy.snakefile | 16 +---------------
 3 files changed, 38 insertions(+), 17 deletions(-)
 create mode 100644 compile-trie.snakefile

diff --git a/README.md b/README.md
index 55048b9..1b55922 100644
--- a/README.md
+++ b/README.md
@@ -28,7 +28,7 @@ This will download the Taxonomy archive from the NCBI FTP server, then unzip the
 
 The download is anonymous and does not require any registration.
 
-### 2. DSMZ catalog download
+### 2. DSMZ catalog download BROKEN
 
 ```shell
 snakemake -j 1 -s dsmz-download.snakefile
@@ -65,7 +65,7 @@ The output of this step contains 5 files:
 snakemake -j 1 -s rewrite-taxonomy.snakefile
 ```
 
-**ETA: 12 minutes**
+**ETA: 20 minutes**
 
 This will write the merged taxonomy in a format suitable for text projection.
 
@@ -79,3 +79,19 @@ This will write the merged taxonomy in a format suitable for text projection.
 | | | `https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=taxonomy&id={ID}` |
 | `bd`   | BacDive | `https://bacdive.dsmz.de/strain/{ID} |
 | | | `https://api.bacdive.dsmz.de/example/fetch/{ID}` (authentication required) |
+
+
+### 5. Compile taxonomy tries
+
+
+```shell
+snakemake -j 1 -s compile-trie.snakefile
+```
+
+**ETA: 8 minutes**
+
+This will compile taxonomy tries for future use with AlvisNLP.
+
+The output of this step contains 2 files:
+* `taxa+id_full.trie`
+* `taxa+id_microorganisms.trie`
diff --git a/compile-trie.snakefile b/compile-trie.snakefile
new file mode 100644
index 0000000..e3465e3
--- /dev/null
+++ b/compile-trie.snakefile
@@ -0,0 +1,19 @@
+configfile: 'config.yaml'
+
+
+rule all:
+    input:
+        config['OUTDIR'] + '/taxa+id_microorganisms.trie',
+        config['OUTDIR'] + '/taxa+id_full.trie'
+
+
+rule compile:
+    input:
+        config['OUTDIR'] + '/taxa+id_{root}.txt'
+
+    output:
+        config['OUTDIR'] + '/taxa+id_{root}.trie'
+
+
+    shell:
+        '''{config[ALVISNLP]} -J-Xmx24G -alias taxo {input} -alias trie {output} compile-taxonomy.plan'''
diff --git a/rewrite-taxonomy.snakefile b/rewrite-taxonomy.snakefile
index 329f4f8..19feac3 100644
--- a/rewrite-taxonomy.snakefile
+++ b/rewrite-taxonomy.snakefile
@@ -5,22 +5,8 @@ rule all:
     input:
         config['OUTDIR'] + '/taxid_microorganisms.txt',
         config['OUTDIR'] + '/taxa+id_microorganisms.txt',
-        config['OUTDIR'] + '/taxa+id_microorganisms.trie',
         config['OUTDIR'] + '/taxid_full.txt',
-        config['OUTDIR'] + '/taxa+id_full.txt',
-        config['OUTDIR'] + '/taxa+id_full.trie'
-
-
-rule check:
-    input:
-        config['OUTDIR'] + '/taxa+id_{root}.txt'
-
-    output:
-        config['OUTDIR'] + '/taxa+id_{root}.trie'
-
-
-    shell:
-        '''{config[ALVISNLP]} -J-Xmx24G -alias taxo {input} -alias trie {output} compile-taxonomy.plan'''
+        config['OUTDIR'] + '/taxa+id_full.txt'
 
 
 rule microorganisms:
-- 
GitLab