-
Notifications
You must be signed in to change notification settings - Fork 1
/
2_gdsc+tcga.lyx
3946 lines (2944 loc) · 76.9 KB
/
2_gdsc+tcga.lyx
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
#LyX 2.2 created this file. For more info see http://www.lyx.org/
\lyxformat 508
\begin_document
\begin_header
\save_transient_properties true
\origin unavailable
\textclass article
\begin_preamble
\usepackage[font=small,labelfont=bf]{caption}
\newcommand*\name[1]{#1}
\newcommand*\abbrv[1]{#1}
\newcommand*\rpkg[1]{\textit{#1}}
\newcommand*\file[1]{\textit{#1}}
\newcommand{\code}[1]{\texttt{#1}}
\newcommand*\protein[1]{#1}
\newcommand*\gene[1]{\textit{#1}}
\end_preamble
\use_default_options true
\maintain_unincluded_children false
\language british
\language_package default
\inputencoding auto
\fontencoding global
\font_roman "default" "default"
\font_sans "default" "default"
\font_typewriter "default" "default"
\font_math "auto" "auto"
\font_default_family default
\use_non_tex_fonts false
\font_sc false
\font_osf false
\font_sf_scale 100 100
\font_tt_scale 100 100
\graphics default
\default_output_format default
\output_sync 0
\bibtex_command default
\index_command default
\paperfontsize default
\spacing single
\use_hyperref false
\papersize default
\use_geometry false
\use_package amsmath 1
\use_package amssymb 1
\use_package cancel 1
\use_package esint 1
\use_package mathdots 1
\use_package mathtools 1
\use_package mhchem 1
\use_package stackrel 1
\use_package stmaryrd 1
\use_package undertilde 1
\cite_engine basic
\cite_engine_type default
\biblio_style plain
\use_bibtopic false
\use_indices false
\paperorientation portrait
\suppress_date false
\justification true
\use_refstyle 1
\index Index
\shortcut idx
\color #008000
\end_index
\secnumdepth 3
\tocdepth 3
\paragraph_separation indent
\paragraph_indentation default
\quotes_language english
\papercolumns 1
\papersides 1
\paperpagestyle default
\tracking_changes false
\output_changes false
\html_math_output 0
\html_css_as_file 0
\html_be_strict false
\end_header
\begin_body
\begin_layout Part
Gene set methods for drug response
\end_layout
\begin_layout Standard
Pathway methods are often used in a cancer context, both for cell lines
and primary tumours.
Most of the time, the method of choice is to take a gene set from either
Gene Ontology (GO)
\begin_inset CommandInset citation
LatexCommand cite
key "Ashburner2000-xl"
\end_inset
, KEGG
\begin_inset CommandInset citation
LatexCommand cite
key "Kanehisa2000-mp"
\end_inset
or Reactome
\begin_inset CommandInset citation
LatexCommand cite
key "Croft2011-po"
\end_inset
, and calculate a combined expression score using either a Fisher's exact
test (e.g.
by a tool called DAVID
\begin_inset Foot
status open
\begin_layout Plain Layout
note that although this tool is still widely used, it has last been updated
in 2010 and misses a lot of annotations
\begin_inset CommandInset citation
LatexCommand cite
key "Wadi2016-wc"
\end_inset
\end_layout
\end_inset
) if one is to test gene sets against differentially expressed genes, or
some variant of Gene Set Enrichment Analysis (GSEA)
\begin_inset CommandInset citation
LatexCommand cite
key "Subramanian2005-pd"
\end_inset
if the sets are pre-defined, but one wants to avoid cutting of continuous
expression values at an arbitrary threshold.
There are, however, more advanced pathway methods available.
Signalling Pathway Impact Analysis
\begin_inset CommandInset citation
LatexCommand cite
key "Tarca2008-ey"
\end_inset
and Differential Expression Analysis for Pathways
\begin_inset CommandInset citation
LatexCommand cite
key "Haynes2013-eo"
\end_inset
take into account the directionality and sign of edges in a pathway.
Pathifier
\begin_inset CommandInset citation
LatexCommand cite
key "Drier2013-vn"
\end_inset
calculates probable information flow between the set items.
PARADIGM
\begin_inset CommandInset citation
LatexCommand cite
key "Vaske2010-zn"
\end_inset
employs a Bayesian framework that models translation, activity, and interaction
s.
I will leave the more complex methods for a later chapter and focus on
GSEA using different gene sets here.
\end_layout
\begin_layout Standard
GSEA using GO gene sets is ubiquitous, often following a differential expression
analysis to see which higher-level function the differentially expressed
genes mediate.
After computing the enrichment scores, our list of genes is condensed down
to a list of significantly enriched GO categories that may be related to
the phenotype we are observing.
This may work very well in some cases.
There are, however, a couple of caveats to observe: (1) a gene does not
exclusively belong to one process; we might very well get a significant
p-value only caused by the overlap between different sets, (2) if we test
all categories and correct by false discovery rate we might dilute our
signal so much that small categories can no longer be significant, or (3)
the process that did indeed cause our phenotype does not correspond to
a gene set at all (this can be due to missing biological knowledge, annotation
errors, or simply the fact that curators have not yet added a certain gene
to a certain category).
Maybe the most dangerous caveat of them all is that once we see our list
of resulting categories, we are inclined to pick out category that
\begin_inset Quotes eld
\end_inset
makes sense
\begin_inset Quotes erd
\end_inset
.
Taking this selection of desired categories on its head, we may also be
inclined to overlook a category that we don't want to see, e.g.
because the involved process is already known in literature and we could
not publish our new findings in a high-impact journal.
The aim of this chapter is to illustrate these issues.
\end_layout
\begin_layout Standard
I will use this chapter to examine which processes are involved in making
cancer cell lines sensitive or resistant to different drugs in the GDSC
panel
\begin_inset CommandInset citation
LatexCommand cite
key "Garnett2012-dk,Iorio2016-gh"
\end_inset
.
I will not filter the gene sets I use, to see how well represented signalling
pathways are among the top hits for drug sensitivity, where they are known
to play a pivotal role for targeted therapies
\begin_inset CommandInset citation
LatexCommand cite
key "Garnett2012-dk,Iorio2016-gh,Yap2012-mi"
\end_inset
.
\end_layout
\begin_layout Standard
\begin_inset VSpace defskip
\end_inset
\end_layout
\begin_layout Standard
Results obtained in section
\begin_inset CommandInset ref
LatexCommand ref
reference "sec:Pathway-responsive-genes-SPEED"
\end_inset
contributed the pathway scores for latest publication of the GDSC screening.
All analyses, plots, and written text in this thesis I produced myself:
\end_layout
\begin_layout Standard
\begin_inset VSpace defskip
\end_inset
\end_layout
\begin_layout Standard
Iorio F, Knijnenburg TA, Vis DJ, Bignell GR, Menden MP,
\series bold
Schubert M
\series default
, Aben N, Gonçalves E, Barthorpe S, Lightfoot H, Cokelaer T, Greninger P,
van Dyk E, Chang H, de Silva H, Heyn H, Deng X, Egan RK, Liu Q, Mironenko
T, Mitropoulos X, Richardson L, Wang J, Zhang T, Moran S, Sayols S, Soleimani
M, Tamborero D, Lopez-Bigas N, Ross-Macdonald P, Esteller M, Gray NS, Haber
DA, Stratton MR, Benes CH, Wessels LF, Saez-Rodriguez J, McDermott U, Garnett
MJ.
\begin_inset Quotes eld
\end_inset
\shape italic
A Landscape of Pharmacogenomic Interactions in Cancer
\shape default
\begin_inset Quotes erd
\end_inset
.
\series bold
Cell
\series default
(2016).
\end_layout
\begin_layout Section
Methods used throughout this thesis
\end_layout
\begin_layout Subsection
Gene sets
\end_layout
\begin_layout Standard
To obtain Gene Ontology sets, I used the
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
rpkg{BioMart}
\end_layout
\end_inset
R package
\begin_inset CommandInset citation
LatexCommand cite
key "Smedley2009-xo"
\end_inset
to query the Ensembl
\begin_inset CommandInset citation
LatexCommand cite
key "Hubbard2002-lv,Yates2016-bw"
\end_inset
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
code{hsapiens
\backslash
_gene
\backslash
_ensembl}
\end_layout
\end_inset
database for all
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
abbrv{HGNC}
\end_layout
\end_inset
symbols that had a Gene Ontology
\begin_inset CommandInset citation
LatexCommand cite
key "Ashburner2000-xl,Gene_Ontology_Consortium2004-sn"
\end_inset
ID (
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
code{go
\backslash
_id}
\end_layout
\end_inset
field) associated with them, yielding three main categories (biological
process, molecular function, cellular compartment) with 16413 gene sets
covering 18806 genes total
\begin_inset Foot
status open
\begin_layout Plain Layout
query of Ensembl Biomart on March 1st 2016
\end_layout
\end_inset
.
For Reactome
\begin_inset CommandInset citation
LatexCommand cite
key "Croft2011-po"
\end_inset
, I downloaded the file
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
file{ReactomePathways.gmt}
\end_layout
\end_inset
\begin_inset Foot
status open
\begin_layout Plain Layout
\begin_inset CommandInset href
LatexCommand href
target "http://www.reactome.org/pages/download-data/"
\end_inset
\end_layout
\end_inset
.
It contained a total of 1675 pathways covering 7852 genes.
\end_layout
\begin_layout Standard
For other gene sets, I used the
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
name{Enrichr}
\end_layout
\end_inset
platform
\begin_inset CommandInset citation
LatexCommand cite
key "Chen2013-da"
\end_inset
and the gene sets the authors assembled in their GitHub repository
\begin_inset Foot
status open
\begin_layout Plain Layout
\begin_inset CommandInset href
LatexCommand href
target "https://github.com/yokuyuki/Enrichr"
\end_inset
\end_layout
\end_inset
.
They encompassed gene sets for 35 pathway and pathway-related resources,
including Gene Ontology, Reactome (where I queried the original databases
to obtain more up-to-date gene lists), as well as KEGG
\begin_inset CommandInset citation
LatexCommand cite
key "Kanehisa2000-mp"
\end_inset
(that already used the last non-commercial release).
\end_layout
\begin_layout Subsection
Gene Set Variation Analysis (GSVA)
\begin_inset CommandInset label
LatexCommand label
name "subsec:GSVA"
\end_inset
\end_layout
\begin_layout Standard
\begin_inset Float figure
wide false
sideways false
status open
\begin_layout Plain Layout
\align center
\begin_inset Graphics
filename figures/GSEA_GSVA_schema.pdf
width 100text%
\end_inset
\end_layout
\begin_layout Plain Layout
\begin_inset Caption Standard
\begin_layout Plain Layout
\begin_inset Argument 1
status collapsed
\begin_layout Plain Layout
Calculation of the running sum statistic for GSEA and GSVA
\end_layout
\end_inset
Calculation of the running sum statistic for GSEA and GSVA.
Calculation of the running sum statistic (top left) and absolute deviation
for enrichment score in case of GSEA vs.
the difference in GSVA.
Genes are ordered by differential expression, genes that are in the query
set are indicated by black bars.
The red line indicates the running sum score where a score is added each
time there is a hit and subtracted otherwise.
GSEA hence produces a bimodal distribution of scores (left), while GSVA
produces a unimodal distribution (right).
This is why the former needs label shuffling of two conditions (bottom
left) to compute empirical p-values, while the latter produces scores for
each sample (but no statistical significance; bottom right).
\begin_inset CommandInset label
LatexCommand label
name "fig:GSEA_schema"
\end_inset
\end_layout
\end_inset
\end_layout
\end_inset
\end_layout
\begin_layout Standard
Gene Set Enrichment Analysis
\begin_inset CommandInset citation
LatexCommand cite
key "Subramanian2005-pd"
\end_inset
is the
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
latin{de facto}
\end_layout
\end_inset
standard to compute the expression level of a set of genes.
It uses as an input a ranked list of genes (
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
latin{e.g.}
\end_layout
\end_inset
fold changes).
It then computes the running sum of a set of interest by starting at the
beginning of this list and adding a score if the current gene is in the
set, or subtracts a score otherwise.
This can be summarised like the following:
\end_layout
\begin_layout Standard
\begin_inset Formula
\[
s_{g+1}=s_{g}+\left\{ \begin{array}{c}
1/n_{set}\\
-1/(n-n_{set})
\end{array}\right.\begin{array}{c}
g\in set\\
g\notin set
\end{array}
\]
\end_inset
\end_layout
\begin_layout Standard
A schema of this calculation is shown in figure
\begin_inset CommandInset ref
LatexCommand ref
reference "fig:GSEA_schema"
\end_inset
.
In the case of GSEA, the overall score is the maximal deviation from zero.
As is shown in the example, this leads to a bimodal distribution of scores
when testing different sets or the same set on different samples, because
even if the genes in the set of interest are evenly spread, there will
always be a deviation.
As GSEA is commonly used to compute the significance of enrichment between
two conditions (left panels in figure
\begin_inset CommandInset ref
LatexCommand ref
reference "fig:GSEA_schema"
\end_inset
), this is not a problem: we can obtain the distribution of scores under
the null hypothesis by shuffling the labels of the reference and the samples
we are looking at, and then compute the empirical p-value as quantile of
this distribution.
This, however, also means that we need to compare two conditions in order
to do this reliably.
We can not compute enrichment scores for each individual sample.
\end_layout
\begin_layout Standard
Gene Set Variation Analysis
\begin_inset CommandInset citation
LatexCommand cite
key "Hanzelmann2013-xl"
\end_inset
solves this: instead of taking the maximal deviation, it takes the difference
between maximum positive and negative enrichment score.
This directly yields a unimodal distribution of enrichment scores in different
samples that can hence be used in statistical tests that assume normality
\begin_inset Foot
status open
\begin_layout Plain Layout
raw gene enrichment scores could still be used in nonparametric tests, but
they are usually less powerful
\end_layout
\end_inset
.
As I am interested in correlating one continuous value (drug sensitivity)
with the set enrichment score, using GSVA (and the
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
rpkg{GSVA}
\end_layout
\end_inset
R package) instead of GSEA is the natural choice.
\end_layout
\begin_layout Standard
\begin_inset Note Note
status open
\begin_layout Plain Layout
TODO_SUBSECTION: Linear associations –> describe OLS (maybe I can get away
w/o)
\end_layout
\end_inset
\end_layout
\begin_layout Subsection
Drug associations using the half-maximum inhibitory concentration (IC50)
\begin_inset CommandInset label
LatexCommand label
name "subsec:Drug-associations"
\end_inset
\end_layout
\begin_layout Standard
\begin_inset Float figure
wide false
sideways false
status open
\begin_layout Plain Layout
\align center
\begin_inset Graphics
filename figures/IC50_schema.pdf
width 60col%
\end_inset
\end_layout
\begin_layout Plain Layout
\begin_inset Caption Standard
\begin_layout Plain Layout
\begin_inset Argument 1
status collapsed
\begin_layout Plain Layout
Schema for calculation of
\begin_inset Formula $IC_{50}$
\end_inset
values
\end_layout
\end_inset
Calculation of
\begin_inset Formula $IC_{50}$
\end_inset
values.Calculation of
\begin_inset Formula $IC_{50}$
\end_inset
values.
Cell viability is measured at different drug concentrations and then a
drug response curve is fitted to the data.
The halfway point of viability between the minimum (
\begin_inset Formula $E_{min}$
\end_inset
) and maximum effect (
\begin_inset Formula $E_{max}$
\end_inset
) is the
\begin_inset Formula $IC_{50}$
\end_inset
.
The curve is defined by these three values and the steepness of the slope.
\begin_inset CommandInset label
LatexCommand label
name "fig:IC50schema"
\end_inset
\end_layout
\end_inset
\end_layout
\end_inset
\end_layout
\begin_layout Standard
The original GDSC data set contained different dilutions of drugs that the
cell lines in the panel were subjected to, measuring how much it interfered
with their growth.
Since I have got a pathway score for each cell line, I also need a single
value corresponding to the sensitivity to a given drug.
One way to do this is to measure the growth inhibition at different concentrati
ons, and then fit a dose-response curve to the data points, interpolating
(or extrapolating, if necessary) the concentration at which the half-maximal
inhibition occurred.
This term is referred to the
\begin_inset Formula $IC_{50}$
\end_inset
value, and has already been calculated in
\begin_inset CommandInset citation
LatexCommand cite
key "Iorio2016-gh"
\end_inset
.
The curve to fit is of sigmoid shape (figure
\begin_inset CommandInset ref
LatexCommand ref
reference "fig:IC50schema"
\end_inset
) and has the formula:
\end_layout
\begin_layout Standard
\begin_inset Formula
\[
x=IC_{50}\left(\frac{y-E_{min}}{E_{max}-y}\right)^{-slope}
\]
\end_inset
\end_layout
\begin_layout Standard
I obtained already processed gene expression matrix from the GDSC cell lines
and their fitted
\begin_inset Formula $IC_{50}$
\end_inset
values to 265 public drugs from the GDSC publication
\begin_inset CommandInset citation
LatexCommand cite
key "Iorio2016-gh"
\end_inset
.
I performed a linear regression using the
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
code{lm}
\end_layout
\end_inset
function in R between the gene set score as an independent variable (
\begin_inset Formula $S_{j}$
\end_inset
, where
\begin_inset Formula $j$
\end_inset
corresponds to each different phenotype from
\begin_inset Formula $1$
\end_inset
to
\begin_inset Formula $k$
\end_inset
; this could
\begin_inset ERT
status open
\begin_layout Plain Layout
\backslash
latin{e.g.}
\end_layout
\end_inset
be pathways or the presence of a mutation) and the
\begin_inset Formula $log_{10}$
\end_inset
of the
\begin_inset Formula $IC50$
\end_inset
in micro-molar as the response variable (
\begin_inset Formula $D_{i}$
\end_inset
, where
\begin_inset Formula $i$
\end_inset
is the drug index).
I regressed out the contribution of individual tissues by including it
as a covariate (
\begin_inset Formula $T$
\end_inset
) in the fit.
\end_layout
\begin_layout Standard
\begin_inset Formula
\[
D_{i}\sim T+S_{j}\qquad\forall i\in drugs\;\forall j\in phenotypes
\]
\end_inset
\end_layout
\begin_layout Standard
In other words, for each drug
\begin_inset Formula $D_{i}$
\end_inset
, I fit the following model for all cell lines
\begin_inset Formula $c$
\end_inset
in the GDSC panel.
\end_layout
\begin_layout Standard
\begin_inset Formula
\[
\begin{array}{c}
D_{i}^{c_{1}}\\
D_{i}^{c_{2}}\\
D_{i}^{c_{3}}\\
\vdots
\end{array}\sim\begin{array}{c}
T^{c_{1}}\\
T^{c_{2}}\\
T^{c_{3}}\\
\vdots
\end{array}+\begin{array}{c}
S_{j}^{c_{1}}\\
S_{j}^{c_{2}}\\
S_{j}^{c_{3}}\\
\vdots
\end{array}
\]
\end_inset
\end_layout
\begin_layout Standard
I performed this association between every drug and all gene set scores,
yielding an effect size (how many units of drug response changed per unit
of enrichment score) and p-value for each pair.
I corrected the p-values for each pair using the False Discovery Rate (FDR)
\begin_inset CommandInset citation
LatexCommand cite
key "Benjamini1995-rj"
\end_inset
.
In addition, I performed these associations using each tissue separately:
\end_layout
\begin_layout Standard
\begin_inset Formula
\[
D_{i}\sim S_{j}\qquad\forall i\in drugs\;\forall j\in phenotypes\mid T^{c}=t;\quad\forall t\in tissues
\]
\end_inset
\end_layout
\begin_layout Standard
In this case, I only include cell lines
\begin_inset Formula $c$
\end_inset
whose tissue
\begin_inset Formula $T$
\end_inset