STRING & related databases: Large-scale integration of heterogeneous data

1. Lars Juhl Jensen
STRING & related databases
Large-scale integration of
heterogeneous data

2. interaction networks

3. association networks

4. guilt by association

6. protein networks

7. STRING

8. 9.6 million proteins

9. common foundation

10. Exercise 1
Go to https://string-db.org/
Query for human SORCS2 using the
search by name functionality
Make sure you are in evidence view
(check the buttons below the
network)
Why are there multiple lines
connecting the same to two

11. curated knowledge

12. (what we know)

13. protein complexes

14. 3D structures

16. pathways

17. metabolic pathways

18. Letunic & Bork, Trends in Biochemical Sciences, 2008

19. signaling pathways

20. very incomplete

21. experimental data

22. (what we measured)

23. physical interactions

24. Jensen & Bork, Science, 2008

25. genetic interactions

26. Beyer et al., Nature Reviews Genetics, 2007

27. gene coexpression

28. microarrays

29. RNAseq

30. Exercise 2
(Continue from where exercise 1
ended)
Which types of evidence support the
interaction between SORCS2 and
NGFR?
Click on the interaction to view the
popup, which has buttons linking to
full details

31. predictions

32. (what we infer)

33. genomic context

34. evolution

35. gene fusion

36. Korbel et al., Nature Biotechnology, 2004

37. gene neighborhood

38. Korbel et al., Nature Biotechnology, 2004

39. phylogenetic profiles

40. Korbel et al., Nature Biotechnology, 2004

41. a real example

45. Cell
Cellulosomes
Cellulose

46. complications

47. many databases

48. different formats

49. different identifiers

50. variable quality

51. not comparable

52. not same species

53. hard work

54. parsers

55. mapping files

56. quality scores

57. affinity purification

58. von Mering et al., Nucleic Acids Research, 2005

59. phylogenetic profiles

61. score calibration

62. gold standard

63. von Mering et al., Nucleic Acids Research, 2005

64. implicit weighting by quality

65. common scale

66. homology-based transfer

67. orthologous groups

68. Franceschini et al., Nucleic Acids Research, 2013

69. missing most of the data

70. Exercise 3
(Continue from where exercise 2
ended)
Change the network to the
confidence view
Change the confidence cutoff to 0.15;
any changes in proteins or
interactions shown?
Increase the number of interactors

71. text mining

72. >10 km

73. too much to read

74. exponential growth

75. ~40 seconds per paper

76. computer

77. as smart as a dog

78. teach it specific tricks

81. named entity recognition

82. comprehensive lexicon

83. cyclin dependent kinase 1

84. CDC2

85. orthographic variation

86. expansion rules

87. prefixes and suffixes

88. CDC2

89. hCdc2

90. flexible matching

91. spaces and hyphens

92. cyclin dependent kinase 1

93. cyclin-dependent kinase 1

94. “black list”

95. SDS

96. information extraction

97. co-mentioning

98. counting

99. within documents

100. within paragraphs

101. within sentences

102. scoring scheme

105. score calibration

106. summary

107. association networks

108. heterogeneous data

109. common identifiers

110. quality scores

111. protein networks

112. Szklarczyk et al., Nucleic Acids Research, 2017string-db.org

113. STITCH

114. chemical networks

115. Kuhn et al., Nucleic Acids Research, 2016stitch-db.org

116. COMPARTMENTS

117. subcellular localization

118. Binder et al., Database, 2014compartments.jensenlab.org

119. TISSUES

120. tissue expression

121. tissues.jensenlab.org Santos et al., PeerJ, 2015

122. DISEASES

123. disease associations

124. diseases.jensenlab.org Frankild et al., Methods, 2015

125. Exercise 4
Open https://diseases.jensenlab.org
Search for Parkinson’s disease
What is the strongest associated
gene?
Inspect the underlying text-mining
evidence
Open https://tissues.jensenlab.org