Skip to content

Commit 5ef44b6

Browse files
komstatkomstat
committed
v2.9.1
1 parent a014573 commit 5ef44b6

File tree

2 files changed

+53
-27
lines changed

2 files changed

+53
-27
lines changed

‎README.md‎

Lines changed: 4 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -3,8 +3,8 @@
33
## 日本語
44

55
* 生命情報・DDBJ センター
6-
* 公開日: 2024-06-14
7-
* version: v2.9
6+
* 公開日: 2024-07-04
7+
* version: v2.9.1
88

99
[Bioinformation and DDBJ Center](https://www.ddbj.nig.ac.jp/index-e.html) のデータベースに登録するためのメタデータ XML を生成、チェックするツール。
1010
* [DDBJ Sequence Read Archive (DRA)](https://www.ddbj.nig.ac.jp/dra/submission.html): Submission、Experiment、Run と Analysis (任意) XML を生成・チェックするためのエクセルとスクリプト
@@ -13,6 +13,7 @@
1313

1414
## 履歴
1515

16+
* 2024-07-04: v2.9.1 Sample 属性を全て ATTRIBUTE に格納
1617
* 2024-06-14: v2.9 Sample 属性追加、及び、公開指定の削除
1718
* 2024-05-28: v2.8.1 file bug fix
1819
* 2024-05-23: v2.8 md5 桁数エラー追加
@@ -335,6 +336,7 @@ These files are Excel, container images and tools for generation and validation
335336

336337
## History
337338

339+
* 2024-07-04: v2.9.1 Store all sample attributes in SAMPLE_ATTRIBUTE
338340
* 2024-06-14: v2.9 Sample attributes added and hold/release removed
339341
* 2024-05-28: v2.8.1 file bug fix
340342
* 2024-05-23: v2.8 md5 checksum value digit check added

‎exe/excel2xml_jga‎

Lines changed: 49 additions & 25 deletions
Original file line numberDiff line numberDiff line change
@@ -11,6 +11,7 @@ require 'optparse'
1111
#
1212

1313
# Update history
14+
# 2024-07-04 Record all sample attributes in the sample attribute elements
1415
# 2024-05-17 Change the way of describing Sample attributes
1516
# 2022-12-23 Change handling of submission date
1617
# 2022-12-22 AGD
@@ -352,9 +353,12 @@ for num, line in sample_a
352353
else
353354

354355
# 追加属性があれば
355-
if line[17..-1].size > 0
356-
line[17..-1].each_with_index{|attr, idx|
357-
added_attr_h.store(added_attr_name_a[idx], attr.to_s.strip) if added_attr_name_a[idx] && attr && attr.to_s
356+
unless line[17..-1].all?{|e| e.empty? || e.nil?}
357+
line[17..-1].each_with_index{|attr, idx|
358+
if added_attr_name_a[idx] && attr && attr.to_s && !attr.to_s.empty?
359+
added_attr_h.store(added_attr_name_a[idx], attr.to_s.strip)
360+
raise "Replace the default explanation \"Add additional phenotypes ...\" by an user-defined attribute name." if added_attr_name_a[idx] =~ /Add additional phenotypes/i
361+
end
358362
}
359363
end
360364

@@ -896,19 +900,19 @@ sample_f.puts xml_sample.SAMPLE_SET{|sample_set|
896900
for sam in samples_a
897901

898902
sample_set.SAMPLE("accession" => "", "center_name" => center_name, "alias" => sam[0]){|sample|
899-
sample.TITLE(sam[2])
903+
sample.TITLE(sam[2].to_s.strip)
900904
sample.SAMPLE_NAME{|sample_name|
901905

902906
sample_name.TAXON_ID("9606")
903907
sample_name.SCIENTIFIC_NAME("Homo sapiens")
904908
sample_name.COMMON_NAME("Human")
905909

906-
sample_name.DONOR_ID(sam[4])
910+
sample_name.DONOR_ID(sam[4].to_s.strip)
907911

908912
}
909913

910-
sample.SAMPLE_GROUP_TYPE(sam[12])
911-
sample.DESCRIPTION(sam[3])
914+
sample.SAMPLE_GROUP_TYPE(sam[12].to_s.strip)
915+
sample.DESCRIPTION(sam[3].to_s.strip)
912916

913917
# attributes
914918
sample.SAMPLE_ATTRIBUTES{|sample_attributes|
@@ -919,8 +923,26 @@ sample_f.puts xml_sample.SAMPLE_SET{|sample_set|
919923
sample_attribute.VALUE(sam[1].to_s.strip)
920924
}
921925

926+
# sample title
927+
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
928+
sample_attribute.TAG("sample_title")
929+
sample_attribute.VALUE(sam[2].to_s.strip)
930+
}
931+
932+
# description
933+
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
934+
sample_attribute.TAG("description")
935+
sample_attribute.VALUE(sam[3].to_s.strip)
936+
}
937+
938+
# subject id
939+
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
940+
sample_attribute.TAG("subject_id")
941+
sample_attribute.VALUE(sam[4].to_s.strip)
942+
}
943+
922944
# sex
923-
if sam[5]
945+
if sam[5] && !sam[5].empty?
924946
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
925947
sample_attribute.TAG("sex")
926948
sample_attribute.VALUE(sam[5].strip)
@@ -930,7 +952,7 @@ sample_f.puts xml_sample.SAMPLE_SET{|sample_set|
930952
end
931953

932954
# age
933-
if sam[6]
955+
if sam[6] && !sam[6].empty?
934956
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
935957
sample_attribute.TAG("age")
936958
sample_attribute.VALUE(sam[6].strip)
@@ -940,7 +962,7 @@ sample_f.puts xml_sample.SAMPLE_SET{|sample_set|
940962
end
941963

942964
# residence
943-
if sam[7]
965+
if sam[7] && !sam[7].empty?
944966
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
945967
sample_attribute.TAG("residence")
946968
sample_attribute.VALUE(sam[7].strip)
@@ -950,7 +972,7 @@ sample_f.puts xml_sample.SAMPLE_SET{|sample_set|
950972
end
951973

952974
# birth_location
953-
if sam[8]
975+
if sam[8] && !sam[8].empty?
954976
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
955977
sample_attribute.TAG("birth_location")
956978
sample_attribute.VALUE(sam[8].strip)
@@ -960,7 +982,7 @@ sample_f.puts xml_sample.SAMPLE_SET{|sample_set|
960982
end
961983

962984
# disease
963-
if sam[9]
985+
if sam[9] && !sam[9].empty?
964986
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
965987
sample_attribute.TAG("disease")
966988
sample_attribute.VALUE(sam[9].strip)
@@ -970,7 +992,7 @@ sample_f.puts xml_sample.SAMPLE_SET{|sample_set|
970992
end
971993

972994
# ICD10
973-
if sam[10]
995+
if sam[10] && !sam[10].empty?
974996
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
975997
sample_attribute.TAG("ICD10")
976998
sample_attribute.VALUE(sam[10].strip)
@@ -980,7 +1002,7 @@ sample_f.puts xml_sample.SAMPLE_SET{|sample_set|
9801002
end
9811003

9821004
# collection_date
983-
if sam[11]
1005+
if sam[11] && !sam[11].empty?
9841006
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
9851007
sample_attribute.TAG("collection_date")
9861008
sample_attribute.VALUE(sam[11].strip)
@@ -990,39 +1012,39 @@ sample_f.puts xml_sample.SAMPLE_SET{|sample_set|
9901012
end
9911013

9921014
# affection status
993-
if sam[12]
1015+
if sam[12] && !sam[12].empty?
9941016
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
9951017
sample_attribute.TAG("affection_status")
9961018
sample_attribute.VALUE(sam[12].strip)
9971019
}
9981020
end
9991021

10001022
# tissue
1001-
if sam[13]
1023+
if sam[13] && !sam[13].empty?
10021024
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
10031025
sample_attribute.TAG("tissue")
10041026
sample_attribute.VALUE(sam[13])
10051027
}
10061028
end
10071029

10081030
# population
1009-
if sam[14]
1031+
if sam[14] && !sam[14].empty?
10101032
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
10111033
sample_attribute.TAG("population")
10121034
sample_attribute.VALUE(sam[14])
10131035
}
10141036
end
10151037

10161038
# histological_type
1017-
if sam[15]
1039+
if sam[15] && !sam[15].empty?
10181040
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
10191041
sample_attribute.TAG("histological_type")
10201042
sample_attribute.VALUE(sam[15])
10211043
}
10221044
end
10231045

10241046
# is_tumor
1025-
if sam[16]
1047+
if sam[16] && !sam[16].empty?
10261048
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
10271049
sample_attribute.TAG("is_tumor")
10281050
sample_attribute.VALUE(sam[16])
@@ -1046,18 +1068,20 @@ sample_f.puts xml_sample.SAMPLE_SET{|sample_set|
10461068

10471069
end
10481070

1049-
else
1071+
else # if previous_sample_attrs_flag
10501072

10511073
# added attributes
10521074
sample_added_attr_h = Hash.new
10531075
sample_added_attr_h = sam[17]
10541076
unless sample_added_attr_h.empty?
10551077
sample_added_attr_h.each{|attr_name, attr_value|
1056-
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
1057-
sample_attribute.TAG(attr_name)
1058-
sample_attribute.VALUE(attr_value)
1059-
}
1060-
}
1078+
if attr_name && !attr_name.empty? && attr_value && !attr_value.empty?
1079+
sample_attributes.SAMPLE_ATTRIBUTE{|sample_attribute|
1080+
sample_attribute.TAG(attr_name)
1081+
sample_attribute.VALUE(attr_value)
1082+
}
1083+
end
1084+
}
10611085
end
10621086

10631087
end

0 commit comments

Comments
 (0)