1
00:00:06,075 --> 00:00:09,777
(chairman) So, let's welcome Egon,

2
00:00:09,777 --> 00:00:13,479
who will describe how he is trying

3
00:00:13,479 --> 00:00:17,579
to improve the coverage 
of chemical compounds in Wikidata.

4
00:00:18,289 --> 00:00:19,679
(Egon) Yeah, thank you.

5
00:00:20,689 --> 00:00:22,309
So, let's see...

6
00:00:22,684 --> 00:00:24,920
Oh, this is not right.

7
00:00:25,880 --> 00:00:26,900
Sorry.

8
00:00:31,754 --> 00:00:33,323
They put the wrong slide deck.

9
00:00:40,183 --> 00:00:42,645
(person 1) The one was better than--
(person 2) [inaudible]

10
00:00:42,645 --> 00:00:44,142
(person 1) (laughs)

11
00:00:49,061 --> 00:00:50,281
How about this?

12
00:00:52,633 --> 00:00:54,765
This one actually says WikidataCon.

13
00:00:56,540 --> 00:00:58,476
Slightly different slides.

14
00:01:00,169 --> 00:01:05,559
Okay, so yeah, coverage and correctness,

15
00:01:05,559 --> 00:01:08,091
accuracy, quality, if you like.

16
00:01:09,422 --> 00:01:11,590
And the other thing here

17
00:01:11,590 --> 00:01:14,628
is what makes it different 
from some of the other things

18
00:01:14,628 --> 00:01:17,276
that we've seen at the WikidataCon

19
00:01:17,276 --> 00:01:20,160
is how I do this quality,

20
00:01:20,160 --> 00:01:21,910
and the coverage, actually.

21
00:01:21,910 --> 00:01:25,220
So I'm actually taking advantage here
of my background,

22
00:01:25,220 --> 00:01:27,166
which is in cheminformatics,

23
00:01:27,166 --> 00:01:30,112
which is something 
that we use in our research.

24
00:01:30,329 --> 00:01:32,989
And cheminformatics

25
00:01:32,989 --> 00:01:35,119
is a way to understanding 
the chemical structures,

26
00:01:35,119 --> 00:01:36,578
what we will see in a moment.

27
00:01:36,578 --> 00:01:39,840
We can do things that we cannot do 
with the regular toolsets

28
00:01:39,840 --> 00:01:43,524
that we have, like Shape Expressions,
quality constraints and sorts.

29
00:01:45,857 --> 00:01:49,203
Now, one of the interesting
things of chemistry

30
00:01:49,496 --> 00:01:51,732
is that chemical structures,

31
00:01:52,852 --> 00:01:55,472
are sometimes the same,
sometimes not the same,

32
00:01:55,472 --> 00:01:57,676
depending on what you want to know.

33
00:01:58,096 --> 00:02:00,636
And this slide reflects that a bit,

34
00:02:00,636 --> 00:02:05,006
and what we see here 
is biologically the same compounds

35
00:02:05,006 --> 00:02:07,943
but chemically two different compounds.

36
00:02:08,496 --> 00:02:11,911
But at biological levels 
with the [inaudible],

37
00:02:11,911 --> 00:02:13,576
they are in equilibrium,

38
00:02:13,576 --> 00:02:16,485
and you will not be able 
to really distinguish between them,

39
00:02:16,735 --> 00:02:19,906
unless you're looking 
for a particular type of biology

40
00:02:19,906 --> 00:02:21,911
like reaction mechanisms.

41
00:02:23,209 --> 00:02:26,979
Another interesting thing 
about Wikidata and Wikipedia

42
00:02:26,979 --> 00:02:31,546
is that we have things 
like long-chain fatty acids,

43
00:02:31,546 --> 00:02:34,539
chemical concepts
which are not a specific compound

44
00:02:34,539 --> 00:02:36,754
but actually a class of compounds.

45
00:02:37,347 --> 00:02:42,867
Now, this class can be based 
on similar features in the molecules,

46
00:02:42,867 --> 00:02:46,137
so like in the case 
of the long-chain fatty acids--

47
00:02:46,137 --> 00:02:51,777
they all have a long-chain fatty
and an acid group.

48
00:02:51,777 --> 00:02:54,101
In other cases, there are the classes

49
00:02:54,101 --> 00:02:57,118
based more
on the biological functionality,

50
00:02:57,118 --> 00:03:00,443
like a certain type of inhibitor,
like an ACE inhibitor.

51
00:03:01,762 --> 00:03:05,632
And this introduces 
a lot of interesting things,

52
00:03:05,632 --> 00:03:09,650
partly because of this close link 
with Wikipedias.

53
00:03:10,270 --> 00:03:12,602
And one of the things that we see

54
00:03:12,602 --> 00:03:17,982
is that Wikipedia may have a chembox 
for a particular compound

55
00:03:17,982 --> 00:03:20,467
bur actually be about a compound class,

56
00:03:20,467 --> 00:03:25,317
resulting to a slightly different concept

57
00:03:25,317 --> 00:03:27,419
of what the two things 
are actually meaning

58
00:03:27,419 --> 00:03:29,990
and the sitelink being more complicated.

59
00:03:33,189 --> 00:03:38,899
We need this 
for understanding the biology.

60
00:03:38,899 --> 00:03:42,921
So the research in our group 
is understanding the living cell.

61
00:03:43,440 --> 00:03:47,030
The system's biology here 
described in the biological process

62
00:03:47,030 --> 00:03:50,545
is we have a pathway database 
for that--WikiPathways,

63
00:03:50,545 --> 00:03:54,022
and if we look at the chemistry in there
of the small molecules,

64
00:03:55,659 --> 00:03:59,179
the chemistry is sometimes 
described in a lot of details,

65
00:03:59,179 --> 00:04:00,739
sometimes in less detail,

66
00:04:00,739 --> 00:04:04,219
pretty much like this Wikipedia
Wikidata link that we just had

67
00:04:04,219 --> 00:04:08,429
resulting in basically links
to a lot of different databases

68
00:04:08,429 --> 00:04:10,424
with slightly different focuses.

69
00:04:10,889 --> 00:04:15,349
Some databases like LIPID MAPS
and the Human Metabolome Database,

70
00:04:15,349 --> 00:04:17,581
they are very much focused on the biology,

71
00:04:17,581 --> 00:04:20,743
whereas a database like ChEBI,

72
00:04:20,743 --> 00:04:24,615
that's very much focused 
on the chemical entities.

73
00:04:26,235 --> 00:04:27,905
So we try to breach that,

74
00:04:27,905 --> 00:04:33,452
and that two-three years ago gave us
a very interesting insight

75
00:04:33,452 --> 00:04:35,871
that if you look at the lines here

76
00:04:35,871 --> 00:04:40,152
where in blue we have the total number 
of the small molecules

77
00:04:40,152 --> 00:04:41,882
we have in these Pathways

78
00:04:41,882 --> 00:04:44,902
and the numbers in red 
that we can match to that,

79
00:04:44,902 --> 00:04:48,402
there is this gap, and this gap 
is complicated chemistry.

80
00:04:49,272 --> 00:04:52,428
Also, poignantly, things
missing in Wikidata.

81
00:04:52,779 --> 00:04:55,327
So therefore the need 
for date completeness

82
00:04:55,327 --> 00:04:57,717
and the data quality.

83
00:05:00,486 --> 00:05:02,184
And here we have an example.

84
00:05:02,184 --> 00:05:05,642
This is actually 
a curation report of yesterday,

85
00:05:05,642 --> 00:05:08,392
and these are still things 
that we have in Pathways

86
00:05:08,392 --> 00:05:13,672
but that we do not know 
what the equivalent thing is in Wikidata.

87
00:05:14,081 --> 00:05:17,527
And one of the things here
that I'm picking out here

88
00:05:17,527 --> 00:05:18,932
is strigolactone.

89
00:05:19,117 --> 00:05:21,024
And this is a class of compounds.

90
00:05:22,031 --> 00:05:27,075
So we have that in one of our Pathways,
this particular Pathway over there.

91
00:05:29,212 --> 00:05:32,172
So you start matching this 
to Wikidata and Wikipedia,

92
00:05:32,172 --> 00:05:35,390
and to actually use
for this compound Wikipedia page

93
00:05:35,976 --> 00:05:38,866
with these six structures--

94
00:05:38,866 --> 00:05:42,101
images, name, no links, nothing.

95
00:05:42,101 --> 00:05:43,169
Nothing in Wikipedia,

96
00:05:43,169 --> 00:05:45,561
just this information, 
not machine-readable.

97
00:05:47,942 --> 00:05:53,022
So based on the name, 
I can actually find three out of the six

98
00:05:53,022 --> 00:05:54,867
of these compounds in Wikidata,

99
00:05:54,867 --> 00:05:56,642
not linked, not classified.

100
00:05:57,253 --> 00:06:00,433
So if we look at the class 
of strigolactones,

101
00:06:00,433 --> 00:06:03,015
of which these six are examples,

102
00:06:03,015 --> 00:06:05,305
Wikidata did not give us anything.

103
00:06:05,305 --> 00:06:08,070
So that's the kind of curation
that I'm interested in.

104
00:06:09,170 --> 00:06:12,900
On the right here--
that page is actually pretty much empty,

105
00:06:12,900 --> 00:06:17,178
but it's exactly what Scholia is showing
for this class of chemical compounds.

106
00:06:18,315 --> 00:06:22,757
So Scholia is one of the tools 
that I've been using to do this curation.

107
00:06:26,339 --> 00:06:30,548
This missing classification
is a bit of information

108
00:06:30,548 --> 00:06:32,057
missing in Wikidata,

109
00:06:32,057 --> 00:06:34,797
but we can add this classification,

110
00:06:34,797 --> 00:06:36,957
and we can retrieve that 
from some sources.

111
00:06:36,957 --> 00:06:38,687
We will see with LIPID MAPS later,

112
00:06:38,687 --> 00:06:41,157
we can automate 
adding these missing links,

113
00:06:41,660 --> 00:06:44,371
if we understand the chemistry.

114
00:06:46,274 --> 00:06:50,554
So this diagram over here--
we have fatty acid over there again

115
00:06:50,554 --> 00:06:54,052
and the long-chain fatty acid over here

116
00:06:54,052 --> 00:06:56,672
that we saw on one
of the previous slides--

117
00:06:56,672 --> 00:07:00,998
very long-chain fatty acids 
and a number of other fatty acids.

118
00:07:01,336 --> 00:07:04,495
This kind of information helps us see

119
00:07:04,495 --> 00:07:09,325
a [inaudible] of the chemistry 
in Wikidata.

120
00:07:10,351 --> 00:07:13,545
Scholia can visualize the 2D structure,

121
00:07:13,545 --> 00:07:15,998
and this thing is actually 
automatically generated

122
00:07:15,998 --> 00:07:18,701
from the chemical structure in Wikidata

123
00:07:18,948 --> 00:07:22,198
on the fly creating 
in the Scalable Vector Graphics.

124
00:07:22,734 --> 00:07:23,864
(coughing)

125
00:07:24,530 --> 00:07:25,563
Sorry.

126
00:07:25,563 --> 00:07:28,150
With the stereochemistry annotation there

127
00:07:28,150 --> 00:07:32,600
to help the chemist see 
the completeness of the data

128
00:07:32,600 --> 00:07:35,819
because also the stereochemistry 
might be missing.

129
00:07:36,452 --> 00:07:40,402
We also get an overview on Scholia 
of related compounds

130
00:07:40,402 --> 00:07:41,872
based on the InChIKey,

131
00:07:41,872 --> 00:07:45,662
where the first block basically indicates
how the atoms are connected

132
00:07:45,905 --> 00:07:50,325
and the second column 
indicates things like stereochemistry

133
00:07:50,325 --> 00:07:53,838
and things like 
which isotopes are in there,

134
00:07:54,334 --> 00:07:57,334
for example C11 instead of C12

135
00:07:57,334 --> 00:08:01,974
or C13 instead of C12.

136
00:08:03,441 --> 00:08:06,459
The last number 
of the last letter over here

137
00:08:06,459 --> 00:08:08,671
actually indicates the charge,

138
00:08:08,671 --> 00:08:13,868
so that's the example that we saw earlier
between the citric acid and the citrate,

139
00:08:14,493 --> 00:08:16,641
or was it the acetic acid and the acetate?

140
00:08:20,496 --> 00:08:23,086
By putting in a bit of the main knowledge,

141
00:08:23,086 --> 00:08:27,708
we can do a lot more... making sense 
of what we have in Wikidata.

142
00:08:29,463 --> 00:08:32,643
A bit more about Scholia

143
00:08:32,643 --> 00:08:36,373
is that about data completeness 
with the physical and chemical properties,

144
00:08:36,373 --> 00:08:40,070
the literature, those are whole things 
that we want to have access to.

145
00:08:40,070 --> 00:08:45,317
But it only works if we can find 
the right chemical in Wikidata.

146
00:08:48,897 --> 00:08:52,467
We started using Wikidata 
in a number of our projects,

147
00:08:52,467 --> 00:08:54,683
so WikiPathways was one of them.

148
00:08:54,683 --> 00:08:56,043
This is another project

149
00:08:56,043 --> 00:08:59,573
in the area of the nanosafety, 
risk assessment,

150
00:08:59,573 --> 00:09:02,728
where they use OECD testing guidelines

151
00:09:02,728 --> 00:09:07,662
and using Wikidata here 
to make an overview of the experiments.

152
00:09:07,662 --> 00:09:11,762
And this means that we can now
actually start annotating articles

153
00:09:11,762 --> 00:09:14,443
where these protocols have been used.

154
00:09:15,663 --> 00:09:20,389
And in this way, we get a better insight 
in the quality of literature as well.

155
00:09:20,895 --> 00:09:23,742
We get to see which DDTs

156
00:09:23,742 --> 00:09:27,268
are well tested, established 
experimental methods

157
00:09:27,268 --> 00:09:32,193
and an indication of how good the data is 
that came out of that.

158
00:09:35,299 --> 00:09:38,199
Another example--this is nanomaterials,

159
00:09:38,199 --> 00:09:39,922
specific nanomaterials,

160
00:09:40,702 --> 00:09:43,402
where there is a unique code-- 
we've added that--

161
00:09:43,402 --> 00:09:45,277
with the same purpose of being able

162
00:09:45,277 --> 00:09:48,006
to track down literature 
about these nanomaterials.

163
00:09:48,655 --> 00:09:51,913
But, again, we need exact descriptions.

164
00:09:53,101 --> 00:09:56,241
Now, this is 
the LIPID MAPS classification,

165
00:09:56,241 --> 00:09:58,011
and here we see an interesting thing,

166
00:09:58,011 --> 00:10:01,986
and this has shown up 
in some of the presentations,

167
00:10:02,876 --> 00:10:04,236
elsewhere as well.

168
00:10:04,236 --> 00:10:08,109
This idea that some of the things 
that we have in Wikidata

169
00:10:08,109 --> 00:10:10,309
is not always matching the sources.

170
00:10:10,309 --> 00:10:12,055
So different ontological models,

171
00:10:12,055 --> 00:10:16,038
different ideas 
of what a particular thing means.

172
00:10:16,225 --> 00:10:20,955
And so, if we look at the LIPID MAPS,
we have a lipid in the middle

173
00:10:20,955 --> 00:10:22,745
and then a number of classes,

174
00:10:22,745 --> 00:10:25,022
and many of these are in Wikidata.

175
00:10:25,454 --> 00:10:30,534
But here, around actually 
fatty acids or fatty acyls,

176
00:10:30,864 --> 00:10:33,190
that's where there is a mismatch

177
00:10:33,468 --> 00:10:37,288
causing something that should be 
actually purely hierarchical,

178
00:10:37,288 --> 00:10:41,070
actually it started to show 
some loops over there,

179
00:10:41,496 --> 00:10:46,189
the mismatch of two representations 
of a lipid chemistry.

180
00:10:48,636 --> 00:10:51,546
Now, the goal of this work
is not so much to reconcile this

181
00:10:51,546 --> 00:10:53,239
but to visualize it

182
00:10:53,239 --> 00:10:55,809
so that we can understand 
what is going on

183
00:10:55,809 --> 00:10:58,877
and correct things 
that are actually clearly wrong.

184
00:11:06,444 --> 00:11:09,619
The interesting about LIPID MAPS 
is actually that the classification

185
00:11:10,105 --> 00:11:12,848
is indicated in the external identifier.

186
00:11:12,848 --> 00:11:14,605
So one of the things 
that we've been using

187
00:11:14,605 --> 00:11:18,845
is these external numbers
to make this automatic classification

188
00:11:18,845 --> 00:11:22,420
because everything 
that starts with an LMFA05

189
00:11:22,600 --> 00:11:25,909
is actually a fatty alcohol.

190
00:11:26,242 --> 00:11:28,424
So I can translate that 
into Quickstatements,

191
00:11:28,424 --> 00:11:29,982
push that into Quickstatements

192
00:11:29,982 --> 00:11:32,992
and get that annotated in Wikidata.

193
00:11:40,133 --> 00:11:44,223
This slide is just reflecting 
the advantage for LIPID MAPS here

194
00:11:44,223 --> 00:11:46,357
which been collaborating with them

195
00:11:46,357 --> 00:11:50,106
because they get a lot of data
out of Wikidata as well,

196
00:11:50,284 --> 00:11:54,786
which we can cross-reference, 
which we can compare if it's correct.

197
00:11:55,066 --> 00:11:57,973
LIPID MAPS is a quite curated database

198
00:11:58,436 --> 00:12:02,276
but like everyone actually having trouble

199
00:12:02,276 --> 00:12:04,748
with access to literature,

200
00:12:04,748 --> 00:12:07,856
the demand of literature 
and filtering the literature,

201
00:12:07,856 --> 00:12:10,102
getting to the right articles.

202
00:12:12,626 --> 00:12:15,126
Shape Expressions is probably 
something that you've seen.

203
00:12:15,126 --> 00:12:18,546
We have a few of them for chemistry now.

204
00:12:18,546 --> 00:12:21,116
This is the example for racemic mixture.

205
00:12:21,116 --> 00:12:26,236
In a case of racemic mixture,
you want to have two parts in there.

206
00:12:26,236 --> 00:12:27,358
It's a mixture,

207
00:12:27,358 --> 00:12:30,657
so at least two chemical entities 
need to be in there.

208
00:12:31,008 --> 00:12:35,561
Moreover, each of the [inaudible] parts
has to be a chemical compound.

209
00:12:35,561 --> 00:12:41,311
This is another level of a way
we can curate the content.

210
00:12:41,811 --> 00:12:43,755
There have to be more of them.

211
00:12:43,755 --> 00:12:47,775
We have quite a few 
different concepts in Wikidata

212
00:12:47,775 --> 00:12:49,922
like groups of co-compounds.

213
00:12:50,506 --> 00:12:54,710
There is a class that is 
of structurally similar compounds, etc.

214
00:12:57,320 --> 00:13:00,532
If you run a query like this,
this case for the other one,

215
00:13:00,532 --> 00:13:04,245
other schema that we have
for chemical elements,

216
00:13:04,245 --> 00:13:07,059
you can do the same thing-- 
you can run it on a single item

217
00:13:07,059 --> 00:13:10,400
or you can run that on everything 
that is a chemical element.

218
00:13:10,400 --> 00:13:13,800
This is something 
that I can very much recommend

219
00:13:13,800 --> 00:13:16,749
having a look at 
if you have not done so already.

220
00:13:17,877 --> 00:13:21,550
Now, if we go to the automation of things,

221
00:13:22,172 --> 00:13:26,032
here I'm using a tool called Bioclipse.

222
00:13:26,032 --> 00:13:29,222
This is something that we worked on
some time ten years ago.

223
00:13:29,222 --> 00:13:32,246
It's a platform 
for chemistry and biology,

224
00:13:32,246 --> 00:13:34,646
or cheminformatics and bioinformatics.

225
00:13:34,646 --> 00:13:36,746
aimed at automating things,

226
00:13:39,513 --> 00:13:41,453
including visualizations and sorts.

227
00:13:41,453 --> 00:13:45,385
But I've taken that now 
and developed a number of scripts

228
00:13:45,385 --> 00:13:47,345
that I can actually run 
on the command line,

229
00:13:47,345 --> 00:13:50,759
which makes it easier to automate things,
as we will see in a moment,

230
00:13:50,759 --> 00:13:52,193
and doing all sort of things,

231
00:13:52,193 --> 00:13:55,957
for example, the classification 
according to the LIPID MAP identifiers,

232
00:13:56,193 --> 00:14:01,717
that's the scripts all available 
from the GitHub repository here.

233
00:14:02,673 --> 00:14:05,703
And typically, I have them
create Quickstatements

234
00:14:05,703 --> 00:14:08,993
because that gets me 
an additional check step

235
00:14:08,993 --> 00:14:13,872
after I created the Quickstatements 
and see what does data actually look like.

236
00:14:15,979 --> 00:14:18,709
Annotation of main subjects.

237
00:14:18,709 --> 00:14:23,144
This one is my script too, 
starting from SMILES

238
00:14:23,144 --> 00:14:27,473
to actually add chemical compounds
that are not in Wikidata yet,

239
00:14:27,473 --> 00:14:28,818
which happens a lot.

240
00:14:29,678 --> 00:14:34,388
So three or four weeks ago 
I added something like 500 compounds

241
00:14:34,388 --> 00:14:36,358
which our project was looking into

242
00:14:36,358 --> 00:14:41,512
because these are 
volatile compounds in oils.

243
00:14:44,610 --> 00:14:47,130
This script adds the compounds.

244
00:14:47,130 --> 00:14:51,370
They will later on add the annotation 
of which pieces that compound comes from

245
00:14:51,370 --> 00:14:52,870
and what the properties are.

246
00:14:55,914 --> 00:14:59,010
Bioclipse itself is based 
on the Chemistry Development Kit

247
00:14:59,878 --> 00:15:01,868
and a few other libraries.

248
00:15:01,868 --> 00:15:04,723
This allows me to do the chemistry.

249
00:15:04,723 --> 00:15:07,973
And this is a very well-validated toolkit.

250
00:15:07,973 --> 00:15:11,653
The SMILES part has been done 
by John Mayfield.

251
00:15:11,653 --> 00:15:17,105
I have done a lot of validation 
against other tools.

252
00:15:17,522 --> 00:15:20,668
And the quality 
is actually really high now,

253
00:15:21,163 --> 00:15:26,328
comparable or in some cases even better 
of commercial cheminformatics tools.

254
00:15:26,328 --> 00:15:29,283
It has given me a lot of reassurance

255
00:15:29,283 --> 00:15:34,923
that the quality checking that we do 
with this tool on Wikidata

256
00:15:34,923 --> 00:15:37,312
is giving interesting results.

257
00:15:38,197 --> 00:15:40,431
This is the Quickstatements.

258
00:15:40,431 --> 00:15:43,275
Quickstatements
is Magnus' work, of course.

259
00:15:44,379 --> 00:15:48,089
What happens if we take the SMILES,
it calculates the InChI,

260
00:15:48,089 --> 00:15:52,239
and the InChIKey, it even looks up 
based on the InChIKey,

261
00:15:52,239 --> 00:15:54,199
if there is a PubChem identifier

262
00:15:54,199 --> 00:15:58,609
that uses the InChIKey, 
the PubChem identifier,

263
00:15:58,609 --> 00:16:01,255
to see if this compound 
already is in Wikidata.

264
00:16:01,255 --> 00:16:03,780
And only if it's not already there,

265
00:16:03,780 --> 00:16:06,945
then it will actually create
a CREATE statement.

266
00:16:09,330 --> 00:16:12,822
A bit of automatic classification
here is an option.

267
00:16:12,822 --> 00:16:15,162
So if I'm adding a class of compounds,

268
00:16:15,162 --> 00:16:17,980
I can automatically indicate 
what these are all...

269
00:16:17,980 --> 00:16:19,970
this type of compounds,

270
00:16:19,970 --> 00:16:22,599
and I can also indicate, if needed,

271
00:16:22,599 --> 00:16:25,214
if there is a particular article
where I got this information

272
00:16:25,214 --> 00:16:27,709
from automatically adding references.

273
00:16:29,836 --> 00:16:32,796
Well, this is what 
the Quickstatements output looks like

274
00:16:32,796 --> 00:16:35,508
for the annotation of main subjects.

275
00:16:36,277 --> 00:16:38,149
You've probably seen that as well.

276
00:16:40,650 --> 00:16:42,820
A newer thing that I started doing

277
00:16:42,820 --> 00:16:46,003
is actually doing reasoning 
on the data in Wikidata.

278
00:16:46,003 --> 00:16:51,457
So if I have the SMILES, then I can check 
the molecular formula, for example.

279
00:16:51,457 --> 00:16:53,427
I can check the InChIKey.

280
00:16:55,242 --> 00:17:00,498
At some point, what we are going to do
is calculate physicochemical properties

281
00:17:00,498 --> 00:17:03,865
and see if that matches 
what is in Wikidata.

282
00:17:04,520 --> 00:17:07,151
This will highlight typos

283
00:17:07,846 --> 00:17:10,386
or wrong units, for example.

284
00:17:11,301 --> 00:17:14,591
At this moment... 
so this is a run of this morning.

285
00:17:14,591 --> 00:17:17,989
What we see here 
is two tests actually failing,

286
00:17:17,989 --> 00:17:19,539
and this is an example of it.

287
00:17:19,539 --> 00:17:23,300
This is the InChIKey
that is computed from the isomeric SMILES

288
00:17:23,300 --> 00:17:26,957
is different from the InChIKey
given in the entry.

289
00:17:27,700 --> 00:17:32,198
This can result from data 
being pulled in from different resources.

290
00:17:33,011 --> 00:17:35,889
So these are entries, about 300 of them,

291
00:17:35,889 --> 00:17:39,479
in the 160,000 chemicals 
that we have in Wikidata.

292
00:17:39,479 --> 00:17:41,632
So it's a very small amount, really,

293
00:17:42,392 --> 00:17:45,672
where there is information, 
and someone needs to look at it.

294
00:17:47,315 --> 00:17:51,008
Now, these are all organic compounds

295
00:17:51,008 --> 00:17:54,168
and also quite a few inorganic compounds

296
00:17:54,168 --> 00:17:57,281
where these things just work less well.

297
00:17:57,618 --> 00:18:01,105
But I found in the other test 
that is failing

298
00:18:01,105 --> 00:18:04,105
immediately a couple of things 
that are very clearly wrong.

299
00:18:09,186 --> 00:18:11,759
PubChem is a huge database.

300
00:18:11,996 --> 00:18:13,846
They do validation as well.

301
00:18:13,846 --> 00:18:16,622
We are in the process 
of submitting Wikidata there,

302
00:18:16,622 --> 00:18:18,812
which I'm really happy about.

303
00:18:19,350 --> 00:18:22,530
It's in the last validation step 
at this moment.

304
00:18:23,323 --> 00:18:25,643
And this will also mean that PubChem,

305
00:18:25,643 --> 00:18:29,143
which has something 
like 100 million compounds

306
00:18:29,143 --> 00:18:31,149
will actually link back to Wikidata.

307
00:18:31,488 --> 00:18:35,117
It already does this, but via Wikipedia.

308
00:18:35,387 --> 00:18:36,947
(laughing)

309
00:18:36,947 --> 00:18:38,434
Do you recognize it?

310
00:18:38,434 --> 00:18:43,067
With the aforementioned issues there 
of concept mismatches.

311
00:18:43,591 --> 00:18:45,691
So this will give us a second thing.

312
00:18:45,691 --> 00:18:49,679
And there, also, 
using the same Bioclipse scripts

313
00:18:49,679 --> 00:18:51,681
or similar Bioclipse scripts,

314
00:18:51,681 --> 00:18:53,493
we get validation reports,

315
00:18:53,493 --> 00:18:57,111
again indicating things 
that chemists should look at.

316
00:18:59,813 --> 00:19:01,540
That basically wraps it up.

317
00:19:01,803 --> 00:19:04,955
This is still a work in progress,
the article is in preparation.

318
00:19:04,955 --> 00:19:08,048
I've been working 
with Finn here in Scholia

319
00:19:08,048 --> 00:19:10,705
to support this validation.

320
00:19:11,058 --> 00:19:15,538
We're writing up the full work, 
but for now you can look up this poster.

321
00:19:15,538 --> 00:19:19,352
The slides are on the program
of this session,

322
00:19:19,352 --> 00:19:21,825
so you can look at the slides
and look at the details.

323
00:19:22,560 --> 00:19:24,939
And a quick acknowledgment:

324
00:19:24,939 --> 00:19:28,370
some of this work has been done 
by a number of grants that I received.

325
00:19:28,370 --> 00:19:29,836
And thank you very much.

326
00:19:30,282 --> 00:19:32,475
(applause)

327
00:19:35,922 --> 00:19:37,902
(chairman) Are there any questions?

328
00:19:41,142 --> 00:19:42,669
(person 3) Thank you so much for this.

329
00:19:42,669 --> 00:19:44,196
I am [inaudible],

330
00:19:44,196 --> 00:19:47,142
and so far, I've been reading articles

331
00:19:47,142 --> 00:19:49,722
on the [inaudible] Quickipedia 
on different compounds.

332
00:19:49,722 --> 00:19:53,770
I have a little bit more than 70 articles 
with different compounds--

333
00:19:53,770 --> 00:19:55,464
just things I come across.

334
00:19:56,000 --> 00:19:58,021
And my question to you is

335
00:19:58,021 --> 00:20:02,371
if I want to move my chemistry activity 
from Wikipedia to Wikidata,

336
00:20:02,371 --> 00:20:05,300
how can I help
in a way that is very friendly

337
00:20:05,300 --> 00:20:10,031
to somebody who is a beginner 
in that field on Wikidata?

338
00:20:12,262 --> 00:20:15,832
So, if that compound is in Wikipedia and..

339
00:20:15,832 --> 00:20:18,092
Sometimes there is 
actually a Wikidata page.

340
00:20:18,092 --> 00:20:19,972
I occasionally run into this as well,

341
00:20:19,972 --> 00:20:21,902
in the last couple of months 
not so much anymore

342
00:20:21,902 --> 00:20:23,791
but this morning, actually.

343
00:20:25,522 --> 00:20:27,433
And what I typically do then

344
00:20:27,433 --> 00:20:31,097
is I take the SMILES
from [inaudible] infobox

345
00:20:31,097 --> 00:20:32,262
from that compound

346
00:20:32,262 --> 00:20:37,182
or use PubChem to look up the SMILES,
check if the information is complete,

347
00:20:37,182 --> 00:20:38,817
particularly the stereochemistry,

348
00:20:39,282 --> 00:20:43,422
and then I use that 
that creates Wikidata item scripts

349
00:20:43,422 --> 00:20:46,320
to create Quickstatements 
for that compound.

350
00:20:47,506 --> 00:20:50,016
If there already is a Wikidata item,

351
00:20:50,016 --> 00:20:55,916
I basically just update these scripts,

352
00:20:55,916 --> 00:21:00,065
but rather than say, "Create Last,"
I replace the last with the Q-codes

353
00:21:00,065 --> 00:21:01,710
that that item already has.

354
00:21:01,710 --> 00:21:04,650
And then it complements 
or it adds this information

355
00:21:04,650 --> 00:21:06,986
based on the information we had.

356
00:21:07,922 --> 00:21:10,422
This is [manuable],

357
00:21:10,422 --> 00:21:13,677
so you can copy-paste 
a number of SMILES, put it in a file,

358
00:21:13,677 --> 00:21:15,999
and take that.

359
00:21:18,351 --> 00:21:21,671
Extracting that information to Wikidata 
is not something I've automated yet,

360
00:21:21,671 --> 00:21:25,088
but this helps me... 
it's a pretty fast process.

361
00:21:25,782 --> 00:21:28,427
I can show you later 
how to use that software.

362
00:21:30,728 --> 00:21:32,478
(chairman) Are there other questions?

363
00:21:33,515 --> 00:21:34,784
So, I have one.

364
00:21:35,097 --> 00:21:39,825
Do you make an effort 
to, in fact, make this more visible

365
00:21:39,825 --> 00:21:42,445
in this bioinformatics community

366
00:21:42,445 --> 00:21:46,671
so that they can start using
this structured data?

367
00:21:47,566 --> 00:21:49,326
Yeah, I'm actively doing that.

368
00:21:49,326 --> 00:21:52,256
So what I did not mention 
in this presentation so much,

369
00:21:52,256 --> 00:21:58,456
but we saw that in...
I'd have somewhere to start here--

370
00:21:58,456 --> 00:22:00,669
this is an overview 
of different databases.

371
00:22:01,106 --> 00:22:04,706
A similar plot, which actually 
I do not have on this slide deck

372
00:22:04,706 --> 00:22:09,387
is the number of different identifiers
that chemical compounds have,

373
00:22:09,387 --> 00:22:11,347
and I've been working 
with a number of databases,

374
00:22:11,347 --> 00:22:15,622
like MassBank, 
the Environmental Protection Agency,

375
00:22:17,184 --> 00:22:18,987
CompTox Dashboard.

376
00:22:19,507 --> 00:22:21,704
I've added links to the BDB database.

377
00:22:21,704 --> 00:22:24,159
So I'm working with a number of projects

378
00:22:24,159 --> 00:22:27,469
for pulling in additional information,

379
00:22:27,739 --> 00:22:30,030
identifies our links out
to other databases.

380
00:22:31,192 --> 00:22:33,371
Regarding outreach, yes,

381
00:22:33,371 --> 00:22:36,961
so that wrong slide deck
that I was showing at the start,

382
00:22:36,961 --> 00:22:39,283
there was actually 
a presentation two weeks ago

383
00:22:39,283 --> 00:22:42,148
at an Open Science Meeting 
around chemistry.

384
00:22:42,503 --> 00:22:45,843
I'm very much pushing this and...

385
00:22:47,410 --> 00:22:48,951
I see a big future here.

386
00:22:48,951 --> 00:22:50,780
There's a lot of interest.

387
00:22:50,780 --> 00:22:54,590
And making people aware 
of the CC0 license,

388
00:22:54,590 --> 00:22:57,616
that's typically the larger problem.

389
00:22:58,200 --> 00:23:02,967
So we have to pull in
the information carefully.

390
00:23:05,303 --> 00:23:06,713
(chairman) Other questions?

391
00:23:08,643 --> 00:23:10,373
- Okay. 
- Thank you very much.

392
00:23:10,373 --> 00:23:12,036
(chairman) Can we thank the speaker.

393
00:23:12,036 --> 00:23:15,257
(applause)