the algorithm for annotation

[baimingze]

algorithm

1. get annotation info of a word

http://data.bioontology.org/annotator?text=modifications&longest_only=true&whole_word_only=false

2. get the first matched_word ("MODIFICATION") from annotation info (excludes the word not from first char, such as FICATION)

"annotations": [
{
    "from": 1,
    "to": 12,
    "matchType": "SYN",
    "text": "MODIFICATION"

}

3. deal the urls which comes from bioontology.org and the match is as same as the matched_word in step 2, to get the detail infos(Accession, ontology)

"@id": "http://purl.bioontology.org/ontology/HL7/C1554963",
"@id": "http://purl.bioontology.org/ontology/LNC/LA11133-8",

4. collect the synonyms from these urls

results(only with two ontologies: MESH and MS)

synonyms are in the square brackets

samples
['sample']
study
analysis
['analysis', 'chemical analysis', 'assay', 'determination']
sequencing
genome
['genome', 'genomes']
cell
['cell']
human
['human', 'homo sapiens', 'man (taxonomy)', 'man, modern', 'modern man']
performed
cancer
['cancer', 'benign neoplasms', 'tumors', 'neoplasm', 'benign neoplasm', 'cancers', 'neoplasm, benign', 'neoplasms, benign', 'neoplasia', 'tumor']
high
['high']
mass
ms
whole
exome
['exome', 'exomes']
dna
['dna', 'dna, double stranded', 'ds-dna', 'double-stranded dna', 'deoxyribonucleic acids', 'deoxyribonucleic acid', 'ds dna', 'dna, double-stranded']
patients
['patients', 'client', 'clients', 'patient']
disease
['disease', 'diseases']
genes
['genes', 'genetic materials', 'cistrons', 'materials, genetic', 'cistron', 'genetic material', 'material, genetic', 'gene']
genetic
['gene', 'genetic materials', 'cistrons', 'materials, genetic', 'cistron', 'genetic material', 'material, genetic']
used
['use']
identified
mutations
['mutations']
identify
proteome
['proteome', 'proteomes']
cells
['cells', 'cell']
project
wide
['wide']
sample
['sample']
part
gene
['gene', 'genetic materials', 'cistrons', 'materials, genetic', 'cistron', 'genetic material', 'material, genetic']
genomic
spectrometry
['spectrometry', 'analysis, spectrum', 'spectroscopy']
including
individuals
http
based
cases
cohort
lc
associated

[baimingze]

annotated words

these words have no annotation info
['study', 'sequencing', 'performed', 'mass', 'ms', 'whole', 'identified', 'identify', 'project', 'part', 'genomic', 'including', 'individuals', 'http', 'based', 'cases', 'cohort', 'lc']

---

{'mergedWords': ['samples', 'sample'], 'label': 'sample', 'frequent': '1320', 'No': '0'}
{'mergedWords': ['analysis'], 'label': 'analysis', 'frequent': '950', 'No': '1'}
{'mergedWords': ['genome'], 'label': 'genome', 'frequent': '828', 'No': '2'}
{'mergedWords': ['cell'], 'label': 'cell', 'frequent': '1004', 'No': '3'}
{'mergedWords': ['human'], 'label': 'human', 'frequent': '642', 'No': '4'}
{'mergedWords': ['cancer'], 'label': 'cancer', 'frequent': '545', 'No': '5'}
{'mergedWords': ['high'], 'label': 'high', 'frequent': '497', 'No': '6'}
{'mergedWords': ['exome'], 'label': 'exome', 'frequent': '455', 'No': '7'}
{'mergedWords': ['dna'], 'label': 'dna', 'frequent': '446', 'No': '8'}
{'mergedWords': ['patients'], 'label': 'patients', 'frequent': '427', 'No': '9'}
{'mergedWords': ['disease'], 'label': 'disease', 'frequent': '414', 'No': '10'}
{'mergedWords': ['genes', 'gene', 'genetic'], 'label': 'gene', 'frequent': '1119', 'No': '11'}
{'mergedWords': ['used'], 'label': 'used', 'frequent': '373', 'No': '13'}
{'mergedWords': ['mutations'], 'label': 'mutations', 'frequent': '364', 'No': '14'}
{'mergedWords': ['proteome'], 'label': 'proteome', 'frequent': '336', 'No': '15'}
{'mergedWords': ['wide'], 'label': 'wide', 'frequent': '320', 'No': '16'}
{'mergedWords': ['spectrometry'], 'label': 'spectrometry', 'frequent': '287', 'No': '17'}
{'mergedWords': ['associated'], 'label': 'associated', 'frequent': '269', 'No': '18'}

[ypriverol]

@baimingze this work is fantastic my friend, well done. How are you planning to store this information?, As far as I see we have two options:

• As we discuss before we should do it in the mongoDB database. The java library will allow you to handle the database for inserts, updates and retrieve options. You can try with it, but is really simple.
• We can use these words during the annotation process and put the synonyms into the XML file, it will allow us to search by them, but also work in the similarity score and other metrics using this annotations.

My opinion is that this information should be in the XML for two main reasons:

• We can search by sysnomyms given an incredible power to the search and for the same time been able to search not only but what is complete write in the text.

• We will not change the frequently words part of the web-services because the words will be in the XML-files.

  Apart of the MS and MESH ontologies we should explore other important ontologies such as:

- Experimental Factor Ontology - EFO 
- Tissue BRENDA ontology 
- GO (Gene ontology)

@baimingze we can discuss this issues on thursday.

[baimingze]

So far, we can get annotation info from EFO and TBO, except GO

1. example detail info in EFO : http://www.ebi.ac.uk/efo/EFO_0000244
2. TBO: 404 error at"http://data.bioontology.org/ontologies/BTO/classes/BTO_0000759", which is get from "data.bioontology.org/annotator?ontologies=BTO&text=liver"........we can use the self url

"self": "http://data.bioontology.org/ontologies/BTO/classes/http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FBTO_0000759","

1. GO: got empty list from "http://data.bioontology.org/annotator?ontologies=GO&text=gene"