https://github.com/kyzhouhzau/agac

Supplementary data for Bioinformatics.
https://github.com/kyzhouhzau/agac

Last synced: 2 months ago
JSON representation

Supplementary data for Bioinformatics.

• Host: GitHub
• URL: https://github.com/kyzhouhzau/agac
• Owner: kyzhouhzau
• Created: 2019-01-19T07:21:11.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2019-03-18T03:12:35.000Z (over 6 years ago)
• Last Synced: 2025-06-16T05:11:51.482Z (4 months ago)
• Language: AMPL
• Size: 4.75 MB
• Stars: 3
• Watchers: 0
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
Data can be download from https://www.dropbox.com/s/33ghkmgbrhgwuof/AGAC_bio.zip?dl=0

#Steps

* step1

	> Download all pubmed texts and localize them(functionpart/pubtator29)

* step2

	> Search the database with 2001 genes.(2001.txt)

	> There are 1951 genes with search results.(functionpart/pmid)

	> A total of 479,1972 texts were searched out.(functionpart/pmid) (wc -l pmid/*)

	

* step3

	> Filter text with mutation and epilepsy-related names.

	> 965 genes have related texts.(functionpart/Filtered_version2)

	> There are a total of 10,221 articles.(functionpart/Filtered_version2)

	

* step4

	> Split the sentences and merge the upper and lower sentences, based on the rules.(functionpart/sentence_level_version2)

	> There are 683 genes with related descriptions and 10,366 sentences with related descriptions.(functionpart/sentence_level_version2)

* step5

	> The AGAC corpus is used to train the entity recognition model to extract the features of 10,366 sentences.

	> The classification model was trained with the AGAC corpus and the LOF/GOF/Unknown classification was performed on 10366 sentences.

	> Remove the sentence labeled unknown with 581 sentences left.(NER_classification/predict_result.txt)

* step6

	> Download the drugbank data and localize it to the database.

* step7

	> The genes related to 581 sentences were searched in the drugbank database, and the up-and-down relationship was restricted.(interaction/genelist65.txt)

	> Obtained 118 drugs(interaction/filtered_drugs118.txt) and 34 genes（interaction/filtered_genes34.txt）

	> Obtained a drug genetic relationship of 426 pairs.(interaction/result426.txt)

* step8

	> The result of manual proofreading 426 pairs was 65.49%. Remove the unknown modified LOF/GOF predictions instead.(interaction/result_labeled_0106.txt)

	> There are 309 remaining pairs(interaction)and 29 genes.(interaction/genelist.txt)

	> The artificially corrected gene was searched again in the drugbank.

	> Get 281 relationship pairs.(interaction/drugbank_result1.txt)It contains 28 genes and 112 drugs.(interaction/filtered_drugs.txt)(interaction/filtered_genes.txt)

	

	

Run：

1.python splitfile.py and python index.py

The download pubtator text is preprocessed with splitfile.py and localized to the elasticsearch database using the index.py script.

2.python get_pmid.py

The pmid list of the corresponding gene was obtained by inputting 2001 genes.

3.python get_abstract.py

Use this script to search for relevant texts with mutations and epilepsy as query conditions.

4.python sentence_level.py

The search results are divided according to certain rules (including the logical relationship between the upper and lower sentences, including mutations, including the target gene).

5.AGAC_CRF

* The data after the clause is placed in the folder NER_classification\agac_test

* python convert2bio.py

* bash wapiti_agac.sh(NER part)

* Run Decision_tree.py to get the classification result predict_result.txt

* Manual correction of results.

6.Download drubank and localize

* Drugbank.py parses the drugbank data and localizes it to the database with index.py

* Get_drugs.py searches the local database and compares the predicted_result pair after manual correction