Cancer Core Transcription factor Specificity (CaCTS) is an algorithm to identify candidate MTFs using pan-cancer RNA-sequencing data from The Cancer Genome Atlas. In this page we describe the procedures to perform CaCTS analysis. For more results details, please check our SCIENCE ADVANCES manuscript.
Install all packages in the latest version of R.
Downaload CaCTS package and run the following command in a Terminal
R CMD INSTALL CaCTS_1.0.tar.gzFirst, install the devtools package.
install.packages("devtools")Load the devtools package.
library(devtools)Use install_github as follows: Load the devtools package.
install_github("lawrenson-lab/CaCTS")Load the CaCTS package.
library(CaCTS)## ===================================================================
## ___ ___ _____ ___ _
## | | | | _ | |
## | ___ | | |___ | | | | _
## | __| | | | | |_| || |
## |___ |__| |___ | ___| |___ __|
## | |
## | |
## -------------------------------------------------------------------
## Finding Cancer Core Transcription factor Specificity (CaCTS)
## Version:1.0
## ===================================================================
Basically, CaCTS algorithm requires two main data structure inputs: the expression matrix and the annotation table.
# Exoression data
load("files/TCGA.RNA.Rda")
TCGA.RNA[1:4, 1:4]## gene_symbol gene_id TCGA-OR-A5J1-01A-11R-A29S-07
## 1 ? 100130426 0.0000
## 2 ? 100133144 3.2661
## 3 ? 100134869 3.9385
## 4 ? 10357 149.1350
## TCGA-OR-A5J2-01A-11R-A29S-07
## 1 0.0000
## 2 2.6815
## 3 8.9948
## 4 81.0777
# Annotations
annot.table <- read.delim("files/SuppTable1-34-TCGAID.txt")
colnames(annot.table)[3] <- "sample.id"
colnames(annot.table)[2] <- "group.name"
head(annot.table)## Cancer group.name sample.id
## 1 ACC ACC TCGA-OR-A5J1-01A-11R-A29S-07
## 2 ACC ACC TCGA-OR-A5J2-01A-11R-A29S-07
## 3 ACC ACC TCGA-OR-A5J3-01A-11R-A29S-07
## 4 ACC ACC TCGA-OR-A5J5-01A-11R-A29S-07
## 5 ACC ACC TCGA-OR-A5J6-01A-31R-A29S-07
## 6 ACC ACC TCGA-OR-A5J7-01A-11R-A29S-07
annot.table$group.name <- paste0(annot.table$Cancer, "-", annot.table$group.name)
head(data.frame(table(annot.table$group.name)), 20)## Var1 Freq
## 1 ACC-ACC 78
## 2 BLCA-BLCA 404
## 3 BRCA-BRCA 1083
## 4 CESC-CESC 301
## 5 CHOL-CHOL 36
## 6 COAD-COAD 441
## 7 DLBC-DLBC 48
## 8 ESCA-Esophagus Adenocarcinoma NOS 88
## 9 ESCA-Esophagus Squamous Cell Carcinoma 94
## 10 GBM-GBM 156
## 11 HNSC-HNSC 514
## 12 KICH-KICH 65
## 13 KIRC-KIRC 515
## 14 KIRP-KIRP 285
## 15 LAML-LAML 173
## 16 LGG-LGG 514
## 17 LIHC-LIHC 368
## 18 LUAD-LUAD 510
## 19 LUSC-LUSC 487
## 20 MESO-MESO 87
In this step, from all Pancancer expressed genes, we are selecting only TF genes. Published lists of transcription factors (TF) were retrieved from Saint-André et al., 2016 and Lambert et al., 2018. Merging both lists we created a catalogue of 1,671 unique TFs, of which 1,578 were expressed in the pancan data set.
TF.list = read.delim("files/merged.list.1671.TFs.txt", sep = "\t")
f.TCGA.RNA = TCGA.RNA[which(TCGA.RNA$gene_symbol %in% as.character(TF.list$NameTF)),]
dim(f.TCGA.RNA)## [1] 1578 11071
aux = which(!as.character(TF.list$NameTF) %in% f.TCGA.RNA$gene_symbol)
write.table(TF.list[aux,],file = "data/non-expressed-TFs.txt", quote = F, row.names = F)
rownames(f.TCGA.RNA) <- f.TCGA.RNA$gene_symbol
f.TCGA.RNA <- f.TCGA.RNA[,3:ncol(f.TCGA.RNA)]
f.TCGA.RNA[1:5, 1:4]## TCGA-OR-A5J1-01A-11R-A29S-07 TCGA-OR-A5J2-01A-11R-A29S-07
## ADNP2 361.671 693.732
## ADNP 1452.450 2917.060
## AEBP1 471.182 1758.690
## AEBP2 842.459 1689.440
## AFF3 1782.900 2962.550
## TCGA-OR-A5J3-01A-11R-A29S-07 TCGA-OR-A5J5-01A-11R-A29S-07
## ADNP2 350.1700 329.9770
## ADNP 1506.1500 2087.5300
## AEBP1 251.8150 493.4160
## AEBP2 1224.1100 1317.5800
## AFF3 43.2529 54.9961
dim(f.TCGA.RNA)## [1] 1578 11069
To calculate a Jensen-Shannon Divergence (JSD) score for each TF for each tumor type, we adjusted the normalized expression values, to rescale all values to >0.
delta1 = max(f.TCGA.RNA, na.rm = T) - min(f.TCGA.RNA, na.rm = T)
delta2 = max(f.TCGA.RNA, na.rm = T) - 0
f.TCGA.RNA.rs = (f.TCGA.RNA - min(f.TCGA.RNA, na.rm = T)) * delta1 / delta2
f.TCGA.RNA.rs[1:5, 1:4]## TCGA-OR-A5J1-01A-11R-A29S-07 TCGA-OR-A5J2-01A-11R-A29S-07
## ADNP2 362.5552 694.6164
## ADNP 1453.3349 2917.9460
## AEBP1 472.0662 1759.5752
## AEBP2 843.3435 1690.3251
## AFF3 1783.7852 2963.4360
## TCGA-OR-A5J3-01A-11R-A29S-07 TCGA-OR-A5J5-01A-11R-A29S-07
## ADNP2 351.05415 330.86113
## ADNP 1507.03498 2088.41540
## AEBP1 252.69908 494.30025
## AEBP2 1224.99478 1318.46485
## AFF3 44.13683 55.88004
dim(f.TCGA.RNA.rs)## [1] 1578 11069
Next, for each tumor type, a representative sample was generated by calculating the mean expression value for each TF.
matrix.rep <- prepare_representaive_samples(expr.matrix = f.TCGA.RNA.rs, sample.descr = annot.table, save.file = F)
matrix.rep[1:4, 1:4]## ACC-ACC BLCA-BLCA BRCA-BRCA CESC-CESC
## ADNP2 467.2289 649.4564 765.7106 700.8787
## ADNP 1898.5172 2455.4685 3606.1140 2393.4897
## AEBP1 1011.8737 6894.9661 15561.8556 4121.5774
## AEBP2 1245.3867 845.4508 632.1004 954.5149
cancer = "OV-OV"
res.CaCTS <- run_CaCTS_score(matrix.rep, cancer)##
|
| | 0%
|
| | 1%
|
|= | 1%
|
|= | 2%
|
|== | 2%
|
|== | 3%
|
|== | 4%
|
|=== | 4%
|
|=== | 5%
|
|==== | 5%
|
|==== | 6%
|
|==== | 7%
|
|===== | 7%
|
|===== | 8%
|
|====== | 8%
|
|====== | 9%
|
|====== | 10%
|
|======= | 10%
|
|======= | 11%
|
|======= | 12%
|
|======== | 12%
|
|======== | 13%
|
|========= | 13%
|
|========= | 14%
|
|========= | 15%
|
|========== | 15%
|
|========== | 16%
|
|=========== | 16%
|
|=========== | 17%
|
|=========== | 18%
|
|============ | 18%
|
|============ | 19%
|
|============= | 19%
|
|============= | 20%
|
|============= | 21%
|
|============== | 21%
|
|============== | 22%
|
|=============== | 22%
|
|=============== | 23%
|
|=============== | 24%
|
|================ | 24%
|
|================ | 25%
|
|================= | 25%
|
|================= | 26%
|
|================= | 27%
|
|================== | 27%
|
|================== | 28%
|
|=================== | 29%
|
|=================== | 30%
|
|==================== | 30%
|
|==================== | 31%
|
|===================== | 32%
|
|===================== | 33%
|
|====================== | 33%
|
|====================== | 34%
|
|====================== | 35%
|
|======================= | 35%
|
|======================= | 36%
|
|======================== | 36%
|
|======================== | 37%
|
|======================== | 38%
|
|========================= | 38%
|
|========================= | 39%
|
|========================== | 39%
|
|========================== | 40%
|
|========================== | 41%
|
|=========================== | 41%
|
|=========================== | 42%
|
|============================ | 42%
|
|============================ | 43%
|
|============================ | 44%
|
|============================= | 44%
|
|============================= | 45%
|
|============================== | 45%
|
|============================== | 46%
|
|============================== | 47%
|
|=============================== | 47%
|
|=============================== | 48%
|
|================================ | 48%
|
|================================ | 49%
|
|================================ | 50%
|
|================================= | 50%
|
|================================= | 51%
|
|================================= | 52%
|
|================================== | 52%
|
|================================== | 53%
|
|=================================== | 53%
|
|=================================== | 54%
|
|=================================== | 55%
|
|==================================== | 55%
|
|==================================== | 56%
|
|===================================== | 56%
|
|===================================== | 57%
|
|===================================== | 58%
|
|====================================== | 58%
|
|====================================== | 59%
|
|======================================= | 59%
|
|======================================= | 60%
|
|======================================= | 61%
|
|======================================== | 61%
|
|======================================== | 62%
|
|========================================= | 62%
|
|========================================= | 63%
|
|========================================= | 64%
|
|========================================== | 64%
|
|========================================== | 65%
|
|=========================================== | 65%
|
|=========================================== | 66%
|
|=========================================== | 67%
|
|============================================ | 67%
|
|============================================ | 68%
|
|============================================= | 69%
|
|============================================= | 70%
|
|============================================== | 70%
|
|============================================== | 71%
|
|=============================================== | 72%
|
|=============================================== | 73%
|
|================================================ | 73%
|
|================================================ | 74%
|
|================================================ | 75%
|
|================================================= | 75%
|
|================================================= | 76%
|
|================================================== | 76%
|
|================================================== | 77%
|
|================================================== | 78%
|
|=================================================== | 78%
|
|=================================================== | 79%
|
|==================================================== | 79%
|
|==================================================== | 80%
|
|==================================================== | 81%
|
|===================================================== | 81%
|
|===================================================== | 82%
|
|====================================================== | 82%
|
|====================================================== | 83%
|
|====================================================== | 84%
|
|======================================================= | 84%
|
|======================================================= | 85%
|
|======================================================== | 85%
|
|======================================================== | 86%
|
|======================================================== | 87%
|
|========================================================= | 87%
|
|========================================================= | 88%
|
|========================================================== | 88%
|
|========================================================== | 89%
|
|========================================================== | 90%
|
|=========================================================== | 90%
|
|=========================================================== | 91%
|
|=========================================================== | 92%
|
|============================================================ | 92%
|
|============================================================ | 93%
|
|============================================================= | 93%
|
|============================================================= | 94%
|
|============================================================= | 95%
|
|============================================================== | 95%
|
|============================================================== | 96%
|
|=============================================================== | 96%
|
|=============================================================== | 97%
|
|=============================================================== | 98%
|
|================================================================ | 98%
|
|================================================================ | 99%
|
|=================================================================| 99%
|
|=================================================================| 100%
tail(res.CaCTS)## Name value LogValue
## res.821 SOX17 0.3954705 0.4028859
## res.395 HOXD3 0.3722604 0.4291532
## res.188 EMX2 0.3634673 0.4395346
## res.119 CTCFL 0.3597423 0.4440085
## res.472 LHX1 0.3446310 0.4626457
## res.985 WT1 0.1666036 0.7783157
p = visualize_scores(tf.scores = res.CaCTS, topn = 0, ncol = 1, rep.matrix = matrix.rep, query.name = cancer, filename = paste0(cancer, "-CaCTS-scores.pdf"), w=5, h=5)
p
The final candidate MTF list for each cancer type was defined by considering the intersection of the 5% most highly expressed TFs in each tumor type, and the top 5% TFs when ranked by JSD score.
filtered <- filter_by_expression_rank(rep.matrix = matrix.rep, tf.scores = res.CaCTS, query.name = cancer, pn=0.05, pnE=0.05)## Filtering TFs by expression rank...
## File saved as: /home/abraaodem/Code/MTF/CaCTS/CaCTs_res/OV-OV-CaCTS-scores-expression-filtered.txt
filtered## Name value LogValue JSD.rank Expr.mean Expr.rank
## 12 WT1 0.1666036 0.7783157 1 5364.437 17
## 2 EMX2 0.3634673 0.4395346 4 2874.900 55
## 10 SOX17 0.3954705 0.4028859 6 3036.156 46
## 5 MEIS1 0.4361167 0.3603972 9 4125.429 25
## 1 BHLHE41 0.4867695 0.3126766 16 5811.158 13
## 7 PAX8 0.4944187 0.3059051 17 6817.798 9
## 3 ESR1 0.5004109 0.3006732 18 2811.067 60
## 14 ZNF503 0.5198643 0.2841100 28 2411.554 74
## 4 MECOM 0.5394078 0.2680828 35 3053.090 45
## 11 TGIF2 0.5543643 0.2562047 46 2305.685 77
## 6 NR2F6 0.5624774 0.2498949 58 3193.932 41
## 8 PBX1 0.5696953 0.2443573 64 4244.521 24
## 9 PLSCR1 0.5753966 0.2400327 76 2843.742 58
## 13 ZNF217 0.5759398 0.2396229 78 3122.163 43
To predict the candidate MTFs for the other cancer types, the same procedures on Step 2 can be performed by changin the cancer of interest.
D’Alessio, A.C., Fan, Z.P., Wert, K.J., Baranov, P., Cohen, M.A., Saini, J.S., Cohick, E., Charniga, C., Dadon, D., Hannett, N.M., et al. (2015). A Systematic Approach to Identify Candidate Transcription Factors that Control Cell Identity. Stem Cell Reports 5, 763–775.
Lambert, S.A., Jolma, A., Campitelli, L.F., Das, P.K., Yin, Y., Albu, M., Chen, X., Taipale, J., Hughes, T.R., and Weirauch, M.T. (2018). The human transcription factors. Cell 175, 598–599.
Saint-André, V., Federation, A.J., Lin, C.Y., Abraham, B.J., Reddy, J., Lee, T.I., Bradner, J.E., and Young, R.A. (2016). Models of human core transcriptional regulatory circuitries. Genome Res. 26, 385–396.