From 014e209bb74e25ca294ffd0210c915b0008eee00 Mon Sep 17 00:00:00 2001
From: Robert Bossy <Robert.Bossy@inra.fr>
Date: Thu, 11 Apr 2024 17:29:00 +0200
Subject: [PATCH] Handle MeSH tree
removed mesh-tree from roots config files
mesh-download.snakefile download current MeSH tree
scripts/roots2mesh-tree.py translates a root file into a list of MeSH paths
mesh-microbio.snakefile creates the MeSH path list for microorganisms roots
---
config.yaml | 1 +
mesh-download.snakefile | 26 ++++++++++++++++
mesh-microbio.snakefile | 26 ++++++++++++++++
resources/ON-roots.yaml | 33 --------------------
resources/microorganisms-roots.yaml | 33 --------------------
resources/vecteurs-roots.yaml | 2 --
scripts/roots2mesh-tree.py | 48 +++++++++++++++++++++++++++++
7 files changed, 101 insertions(+), 68 deletions(-)
create mode 100644 mesh-download.snakefile
create mode 100644 mesh-microbio.snakefile
create mode 100755 scripts/roots2mesh-tree.py
diff --git a/config.yaml b/config.yaml
index 743ee70..d1a4272 100644
--- a/config.yaml
+++ b/config.yaml
@@ -15,3 +15,4 @@ NCBI_DIR: 'ncbi-taxonomy'
NCBI_ZIP_URL: 'https://ftp.ncbi.nlm.nih.gov/pub/taxonomy/taxdmp.zip'
EPPO_DIR: 'EPPO'
EPPO_ZIP_URL: 'https://data.eppo.int/files/xmlfull.zip'
+MESH_DIR: 'mesh'
diff --git a/mesh-download.snakefile b/mesh-download.snakefile
new file mode 100644
index 0000000..5a84957
--- /dev/null
+++ b/mesh-download.snakefile
@@ -0,0 +1,26 @@
+configfile: 'config.yaml'
+
+
+import datetime
+
+
+OUTDIR=config['OUTDIR'] + '/' + config['MESH_DIR']
+YEAR=str(datetime.date.today().year)
+
+
+rule all:
+ input:
+ OUTDIR + '/d' + YEAR + '.bin'
+
+
+rule mesh:
+ output:
+ OUTDIR + '/d' + YEAR + '.bin'
+
+ params:
+ year=YEAR
+
+ shell:
+ '''wget -O {output} 'https://nlmpubs.nlm.nih.gov/projects/mesh/MESH_FILES/asciimesh/d{params.year}.bin' '''
+
+
diff --git a/mesh-microbio.snakefile b/mesh-microbio.snakefile
new file mode 100644
index 0000000..8bd3dc2
--- /dev/null
+++ b/mesh-microbio.snakefile
@@ -0,0 +1,26 @@
+configfile: 'config.yaml'
+
+
+import datetime
+
+
+OUTDIR=config['OUTDIR'] + '/' + config['MESH_DIR']
+YEAR=str(datetime.date.today().year)
+
+
+rule all:
+ input:
+ OUTDIR + '/microbio-mesh-terms.txt'
+
+
+rule mesh_microbio:
+ output:
+ OUTDIR + '/microbio-mesh-terms.txt'
+
+ input:
+ script='scripts/roots2mesh-tree.py',
+ mesh=OUTDIR + '/d' + YEAR + '.bin',
+ roots='resources/microorganisms-roots.yaml'
+
+ shell:
+ '''{input.script} {input.mesh} {input.roots} >{output}'''
diff --git a/resources/ON-roots.yaml b/resources/ON-roots.yaml
index a1c93ab..e88c8e7 100644
--- a/resources/ON-roots.yaml
+++ b/resources/ON-roots.yaml
@@ -3,124 +3,91 @@
- name: Acari
taxid: ncbi:6933
- mesh-id: D056893
- mesh-tree: B01.043
name: Alveolata
taxid: ncbi:33630
- mesh-id: D056894
- mesh-tree: B01.046
name: Amoebozoa
taxid: ncbi:554915
- mesh-id: D001105
- mesh-tree: B02
name: Archaea
taxid: ncbi:2157
- mesh-id: D001419
- mesh-tree: B03
name: Bacteria
taxid: ncbi:2
- mesh-id: D000077105
- mesh-tree: B01.650.940.150.511
name: Chlamydomonadales
taxid: ncbi:3042
- mesh-id: D002708
- mesh-tree: B01.650.940.150.469
name: Chlorella
taxid: ncbi:3071
- mesh-id: D056897
- mesh-tree: B01.175
name: Choanoflagellida
taxid: ncbi:28009
- mesh-id: D044785
- mesh-tree: B01.206
name: Cryptophyta
taxid: ncbi:3027
- mesh-id: D058114
- mesh-tree: B01.650.940.800.150.200
name: Desmidiales
taxid: ncbi:131210
- mesh-id: D016828
- mesh-tree: B01.237
name: Diplomonadida
taxid: ncbi:5738
- mesh-id: D056898
- mesh-tree: B01.268
name: Euglenozoa
taxid: ncbi:33682
- mesh-id: D005658
- mesh-tree: B01.300
name: Fungi
taxid: ncbi:4751
- mesh-id: D058108
- mesh-tree: B01.650.232
name: Glaucocystophyceae
taxid: ncbi:38254
- mesh-id: D058087
- mesh-tree: B01.400
name: Haptophyta
taxid: ncbi:2830
- mesh-id: D050298
- mesh-tree: B01.500
name: Ichthyosporea
taxid: ncbi:127916
- mesh-id: D009348
- mesh-tree: B01.050.500.500.294
name: Nematoda
taxid: ncbi:6231
- mesh-id: D056899
- mesh-tree: B01.625
name: Oxymonadida
taxid: ncbi:66288
- mesh-id: D056900
- mesh-tree: B01.630
name: Parabasalia
taxid: ncbi:5719
- mesh-id: D011525
- mesh-tree: B01.650.940.150.634
name: Prototheca
taxid: ncbi:3110
- mesh-id: D056919
- mesh-tree: B01.675
name: Retortamonadidae
taxid: ncbi:193075
- mesh-id: D056901
- mesh-tree: B01.680
name: Rhizaria
taxid: ncbi:543769
- mesh-id: D058009
- mesh-tree: B01.750
name: Stramenopiles
taxid: ncbi:33634
- mesh-id: D014780
- mesh-tree: B04
name: Viruses
taxid: ncbi:10239
- mesh-id: D008827
- mesh-tree: G06
name: Microbiological Phenomena
- mesh-id: D008829
- mesh-tree: H01.158.273.540
name: Microbiology
- mesh-id: D008828
- mesh-tree: E05.200.875,E01.370.225.875
name: Microbiological Techniques
- mesh-id: D001287
- mesh-tree: G05.360.340.024.079
name: Attachment Sites, Microbiological
- mesh-id: D064806
- mesh-tree: C23.550.308
name: Dysbiosis
- mesh-id: D054892
- mesh-tree: G05.360.340.550
name: Metagenome
- mesh-id: D064349
- mesh-tree: G05.360.340.358
name: Genome, Microbial
- mesh-id: D056226
- mesh-tree: A20
name: Bacterial Structures
- mesh-id: D056229
- mesh-tree: A19
name: Fungal Structures
- mesh-id: D056224
- mesh-tree: A21
name: Viral Structures
diff --git a/resources/microorganisms-roots.yaml b/resources/microorganisms-roots.yaml
index 0bbbeb7..9801ba0 100644
--- a/resources/microorganisms-roots.yaml
+++ b/resources/microorganisms-roots.yaml
@@ -1,123 +1,90 @@
- mesh-id: D056893
- mesh-tree: B01.043
name: Alveolata
taxid: ncbi:33630
- mesh-id: D056894
- mesh-tree: B01.046
name: Amoebozoa
taxid: ncbi:554915
- mesh-id: D001105
- mesh-tree: B02
name: Archaea
taxid: ncbi:2157
- mesh-id: D001419
- mesh-tree: B03
name: Bacteria
taxid: ncbi:2
- mesh-id: D000077105
- mesh-tree: B01.650.940.150.511
name: Chlamydomonadales
taxid: ncbi:3042
- mesh-id: D002708
- mesh-tree: B01.650.940.150.469
name: Chlorella
taxid: ncbi:3071
- mesh-id: D056897
- mesh-tree: B01.175
name: Choanoflagellida
taxid: ncbi:28009
- mesh-id: D044785
- mesh-tree: B01.206
name: Cryptophyta
taxid: ncbi:3027
- mesh-id: D058114
- mesh-tree: B01.650.940.800.150.200
name: Desmidiales
taxid: ncbi:131210
- mesh-id: D016828
- mesh-tree: B01.237
name: Diplomonadida
taxid: ncbi:5738
- mesh-id: D056898
- mesh-tree: B01.268
name: Euglenozoa
taxid: ncbi:33682
- mesh-id: D005658
- mesh-tree: B01.300
name: Fungi
taxid: ncbi:4751
- mesh-id: D058108
- mesh-tree: B01.650.232
name: Glaucocystophyceae
taxid: ncbi:38254
- mesh-id: D058087
- mesh-tree: B01.400
name: Haptophyta
taxid: ncbi:2830
- mesh-id: D050298
- mesh-tree: B01.500
name: Ichthyosporea
taxid: ncbi:127916
- mesh-id: D009348
- mesh-tree: B01.050.500.500.294
name: Nematoda
taxid: ncbi:6231
- mesh-id: D056899
- mesh-tree: B01.625
name: Oxymonadida
taxid: ncbi:66288
- mesh-id: D056900
- mesh-tree: B01.630
name: Parabasalia
taxid: ncbi:5719
- mesh-id: D011525
- mesh-tree: B01.650.940.150.634
name: Prototheca
taxid: ncbi:3110
- mesh-id: D056919
- mesh-tree: B01.675
name: Retortamonadidae
taxid: ncbi:193075
- mesh-id: D056901
- mesh-tree: B01.680
name: Rhizaria
taxid: ncbi:543769
- mesh-id: D058009
- mesh-tree: B01.750
name: Stramenopiles
taxid: ncbi:33634
- mesh-id: D014780
- mesh-tree: B04
name: Viruses
taxid: ncbi:10239
- mesh-id: D008827
- mesh-tree: G06
name: Microbiological Phenomena
- mesh-id: D008829
- mesh-tree: H01.158.273.540
name: Microbiology
- mesh-id: D008828
- mesh-tree: E05.200.875,E01.370.225.875
name: Microbiological Techniques
- mesh-id: D001287
- mesh-tree: G05.360.340.024.079
name: Attachment Sites, Microbiological
- mesh-id: D064806
- mesh-tree: C23.550.308
name: Dysbiosis
- mesh-id: D054892
- mesh-tree: G05.360.340.550
name: Metagenome
- mesh-id: D064349
- mesh-tree: G05.360.340.358
name: Genome, Microbial
- mesh-id: D056226
- mesh-tree: A20
name: Bacterial Structures
- mesh-id: D056229
- mesh-tree: A19
name: Fungal Structures
- mesh-id: D056224
- mesh-tree: A21
name: Viral Structures
diff --git a/resources/vecteurs-roots.yaml b/resources/vecteurs-roots.yaml
index 1bc763e..fa358f3 100644
--- a/resources/vecteurs-roots.yaml
+++ b/resources/vecteurs-roots.yaml
@@ -3,10 +3,8 @@
- name: Arachnida
taxid: ncbi:6854
- mesh-id: D009348
- mesh-tree: B01.050.500.500.294
name: Nematoda
taxid: ncbi:6231
- mesh-id: D005658
- mesh-tree: B01.300
name: Fungi
taxid: ncbi:4751
diff --git a/scripts/roots2mesh-tree.py b/scripts/roots2mesh-tree.py
new file mode 100755
index 0000000..8f3d1bf
--- /dev/null
+++ b/scripts/roots2mesh-tree.py
@@ -0,0 +1,48 @@
+#!/bin/env python3
+
+
+import sys
+import yaml
+import re
+import collections
+
+
+MESH_FILE = sys.argv[1]
+ROOTS_FILE = sys.argv[2]
+
+
+with open(ROOTS_FILE) as f:
+ ROOTS = dict((e['mesh-id'], e['name']) for e in yaml.load(f, Loader=yaml.FullLoader) if 'mesh-id' in e)
+
+
+LINE = re.compile('([1A-Z_ ]+) = (.*)')
+
+
+def finish(rec):
+ if 'UI' not in rec:
+ return
+ if rec['UI'][0] not in ROOTS:
+ return
+ if 'MN' not in rec:
+ raise RuntimeError('no tree for ' + rec['UI'][0])
+ for mtree in rec['MN']:
+ print(mtree)
+
+
+current = collections.defaultdict(list)
+with open(MESH_FILE) as f:
+ for line in f:
+ line = line.strip()
+ if line == '':
+ continue
+ if line == '*NEWRECORD':
+ finish(current)
+ current = collections.defaultdict(list)
+ continue
+ m = LINE.fullmatch(line)
+ if m is None:
+ raise RuntimeError('could not parse: ' + line)
+ key = m.group(1)
+ value = m.group(2)
+ current[key].append(value)
+finish(current)
--
GitLab