Large-scale integration of data and text
Download as PPT, PDF2 likes386 views
This document discusses large-scale integration of biological data and text through several databases and resources created by the author's lab, including STRING and STITCH, which integrate protein-protein interaction data from multiple sources. It describes using text mining to extract protein interactions from literature through named entity recognition, relationship extraction, and integration of the results with interaction data from experimental and computational sources. Clinical applications including using Danish health registries and text mining of medical records to discover new drug-drug interactions and adverse drug reactions are also summarized.
![Gene and protein names
Cue words for entity
recognition
Verbs for relation extraction
[nxexpr The expression of
[nxgene the cytochrome
genes
[nxpg CYC1 and CYC7]]]
is controlled by
[nxpg HAP1]
Saric et al., Proceedings of ACL, 2004](https://image.slidesharecdn.com/jensen2015talk5-150415235905-conversion-gate02/85/Large-scale-integration-of-data-and-text-78-320.jpg)
Large-scale integration of data and text
- 1. Large-scale integration of data and text Lars Juhl Jensen
- 3. text mining
- 5. medicine
- 9. STRING
- 10. Szklarczyk et al., Nucleic Acids Research, 2015string-db.org
- 11. 2000+ genomes
- 12. genomic context
- 13. gene fusion
- 14. Korbel et al., Nature Biotechnology, 2004
- 15. operons
- 16. Korbel et al., Nature Biotechnology, 2004
- 18. Korbel et al., Nature Biotechnology, 2004
- 20. Korbel et al., Nature Biotechnology, 2004
- 21. a real example
- 30. Jensen & Bork, Science, 2008
- 32. Beyer et al., Nature Reviews Genetics, 2007
- 34. pathways
- 35. Letunic & Bork, Trends in Biochemical Sciences, 2008
- 36. many databases
- 39. variable quality
- 40. not comparable
- 41. not same species
- 42. hard work
- 43. (Ph.D. students)
- 44. quality scores
- 45. von Mering et al., Nucleic Acids Research, 2005
- 46. calibrate vs. gold standard
- 47. von Mering et al., Nucleic Acids Research, 2005
- 49. Franceschini et al., Nucleic Acids Research, 2013
- 50. missing most of the data
- 51. text mining
- 52. >10 km
- 53. too much to read
- 54. computer
- 55. as smart as a dog
- 56. teach it specific tricks
- 61. cyclin dependent kinase 1
- 62. CDC2
- 64. cyclin dependent kinase 1
- 67. CDC2
- 68. hCdc2
- 69. “black list”
- 70. SDS
- 72. co-mentioning
- 73. within documents
- 75. within sentences
- 78. Gene and protein names Cue words for entity recognition Verbs for relation extraction [nxexpr The expression of [nxgene the cytochrome genes [nxpg CYC1 and CYC7]]] is controlled by [nxpg HAP1] Saric et al., Proceedings of ACL, 2004
- 79. more precise
- 80. worse recall
- 82. STITCH
- 83. STRING + 300k chemicals
- 84. Kuhn et al., Nucleic Acids Research, 2014stitch-db.org
- 85. COMPARTMENTS
- 86. Binder et al., Database, 2014compartments.jensenlab.org
- 87. TISSUES
- 88. tissues.jensenlab.org Santos et al., submitted, 2015
- 89. DISEASES
- 90. diseases.jensenlab.org Frankild et al., Methods, 2015
- 94. text mining
- 97. quality scores
- 98. visualization
- 99. web resources
- 100. download files
- 101. why so many?
- 102. Swiss army knife syndrome
- 104. targeted resources
- 106. medical data mining
- 107. Jensen et al., Nature Reviews Genetics, 2012
- 109. opt-out
- 110. opt-in
- 111. structured data
- 112. Jensen et al., Nature Reviews Genetics, 2012
- 114. established in 1968
- 115. Jensen et al., Nature Reviews Genetics, 2012
- 117. 14 years
- 118. 6.2 million patients
- 120. Jensen et al., Nature Reviews Genetics, 2012
- 121. guilt by association
- 122. naïve approach
- 123. comorbidity
- 124. Jensen et al., Nature Reviews Genetics, 2012
- 125. confounding factors
- 126. “known knowns”
- 127. gender
- 128. age
- 129. type of hospital encounter
- 130. Jensen et al., Nature Communications, 2014
- 131. “known unknowns”
- 132. smoking
- 133. diet
- 134. “unknown unknowns”
- 135. reporting biases
- 136. matched controls
- 138. trajectories
- 139. Jensen et al., Nature Communications, 2014
- 140. trajectory networks
- 141. Jensen et al., Nature Communications, 2014
- 142. complex networks
- 143. key diagnoses
- 144. Jensen et al., Nature Communications, 2014
- 146. medical text mining
- 147. pharmacovigilance
- 148. unstructured data
- 150. Danish
- 152. drugs
- 153. Clozapine
- 155. adverse drug events
- 156. rule-based system
- 157. Eriksson et al., Drug Safety, 2014 Drug introduction Drug discontinuationAdverse event Adverse eventNegative modifier Indication Pre-existing condition Adverse drug reaction Possible adverse drug reaction ADR of additional drug
- 158. Eriksson et al., Drug Safety, 2014 Drug introduction Drug discontinuationAdverse eventIdentification start Adverse eventNegative modifier Indication Pre-existing condition Adverse drug reaction Possible adverse drug reaction ADR of additional drug
- 159. Eriksson et al., Drug Safety, 2014 Drug introduction Drug discontinuation Adverse eventNegative modifier Indication Pre-existing condition Adverse drug reaction Possible adverse drug reaction Adverse event ADR of additional drug Identification start
- 160. Eriksson et al., Drug Safety, 2014 Drug introduction Drug discontinuation Adverse eventNegative modifier Indication Pre-existing condition Adverse drug reaction Possible adverse drug reaction Adverse event ADR of additional drug Identification start
- 161. new adverse drug reactions
- 162. Eriksson et al., Drug Safety, 2014 Drug substance ADE p-value Chlordiazepoxide Nystagmus 4.0e-8 Simvastatin Personality changes 8.4e-8 Dipyridamole Visual impairment 4.4e-4 Citalopram Psychosis 8.8e-4 Bendroflumethiazi de Apoplexy 8.5e-3
- 164. Eriksson et al., Drug Safety, 2014
- 165. Acknowledgments STRING/STITCH Michael Kuhn Damian Szklarczyk Andrea Franceschini Milan Simonovic Alexander Roth Sune Pletscher-Frankild Jianyi Lin Pablo Minguez Christian von Mering Peer Bork Text mining Sune Pletscher- Frankild Jasmin Saric Evangelos Pafilis Alberto Santos Janos Binder Kalliopi Tsafou Heiko Horn Michael Kuhn Reinhardt Schneider Sean O’ Donoghue EHR mining Anders Boeck Jensen Robert Eriksson Peter Bjødstrup Jensen Andreas Bok Andersen Sabrina Gade Ellesøe Henriette Schmock Tudor Oprea Pope Moseley Thomas Werge Søren Brunak