Introduction on how to use Select package to find Evolutionary Dependencies.

Arvind Iyer

2023-08-28

Source: vignettes/introduction.Rmd
  • The goal of select pacakge is to detect Evolutionary Dependecies between alterations/genes in cancer.
  • select package provides function to generate the backgorund model and other utilites functions.

Installation

  • You can install the development version of select from GitHub with:
# install.packages("devtools")
# devtools::install_github("CSOgroup/select")

Example

  • We will run select algorithm on processed LUAD dataset from TCGA provided with the package.
library(select)
library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following object is masked from 'package:select':
#> 
#>     select
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

Data Description & Format

  • The loaded data is list object which consists of
    • gam: a presence absence matrix of alterations
    • samples: a named vector of sample annotations
    • alt: a named vector of alteration annotations
# Check the data strucutre
## Load the data provided with the package
data(luad_data, package = "select")
str(luad_data)
#> List of 3
#>  $ gam    : num [1:502, 1:659] 0 0 0 0 0 0 0 0 1 0 ...
#>   ..- attr(*, "dimnames")=List of 2
#>   .. ..$ : chr [1:502] "TCGA-05-4244-01" "TCGA-05-4249-01" "TCGA-05-4250-01" "TCGA-05-4382-01" ...
#>   .. ..$ : chr [1:659] "MUT.ABL1" "MUT.ACVR1B" "MUT.ACVR2A" "MUT.AKT1" ...
#>  $ alt    : Named chr [1:659] "MUT" "MUT" "MUT" "MUT" ...
#>   ..- attr(*, "names")= chr [1:659] "MUT.ABL1" "MUT.ACVR1B" "MUT.ACVR2A" "MUT.AKT1" ...
#>  $ samples: Named chr [1:502] "LUAD" "LUAD" "LUAD" "LUAD" ...
#>   ..- attr(*, "names")= chr [1:502] "TCGA-05-4244-01" "TCGA-05-4249-01" "TCGA-05-4250-01" "TCGA-05-4382-01" ...

Running select

  • We use the function select() which generates the background model and results.
  • The parameters for the functions are:
    • M: the list object of GAMs & TMB
    • sample.class: a named vector of samples with covariates
    • alteration.class: a named vector of alteration with covariates
    • many more
  • The function returns a dataframe.
alpi <- select::select(
  M = luad_data$gam,
  sample.class = luad_data$samples,
  alteration.class = luad_data$alt,
  folder = './',
  r.seed = 110,
  n.cores = 1,
  vetos = NULL,
  n.permut = 100,
  min.feature.support=5,
  min.feature.freq=0.001,
  remove.0.samples=TRUE,
  remove.unknown.class.samples=TRUE,
  rho = 0.1,
  lambda = 15,
  save.intermediate.files = FALSE,
  randomization.switch.threshold = 30,
  max.memory.size=100,
  calculate_APC_threshold = TRUE,
  calculate_FDR = TRUE,
  verbose = TRUE
  )
#> [1] "Running SELECT... [ .// ]"
#> [1] "-> Parsing and Filtering GAM..."
#> [1] "-> Generating null model..."
#>    user  system elapsed 
#>   2.899   0.052   2.951 
#> [1] "Memory usage: 29.9 MB"
#> [1] "-> Collecting event stats on observed GAM..."
#>    user  system elapsed 
#>   0.686   3.043   0.054 
#> [1] "Memory usage: 30.2 MB"
#> [1] "-> Calculating wMI on observed GAM..."
#>    user  system elapsed 
#>   0.022   0.001   0.023 
#> [1] "Memory usage: 30.6 MB"
#> [1] "-> Collecting event stats on null model..."
#>    user  system elapsed 
#>  34.238 158.332   2.350 
#> [1] "Memory usage: 36.2 MB"
#> [1] "-> Comparing observed GAM vs null model..."
#>    user  system elapsed 
#>   1.186   0.002   1.188 
#> [1] "Memory usage: 36.8 MB"
#> [1] "-> Calculating wMI on null model..."
#>    user  system elapsed 
#>   1.370   0.020   1.391 
#> [1] "Memory usage: 58.8 MB"
#> [1] "-> Calculating P-value based on wMI..."
#>    user  system elapsed 
#>   0.027   0.006   0.034 
#> [1] "Memory usage: 59 MB"
#> [1] "-> Deriving average background wMI..."
#>    user  system elapsed 
#>   0.995   0.002   0.997 
#> [1] "Memory usage: 59.3 MB"
#> [1] "-> Building results table..."
#> [1] "-> Calculating APC score..."
#> [1] "Not correcting with APC as there are not enough interactions to estimate the background distribution. Uncorrected scores will be used."
#> [1] "Not correcting with APC as there are not enough interactions to estimate the background distribution. Uncorrected scores will be used."
#> [1] "Not correcting with APC as there are not enough interactions to estimate the background distribution. Uncorrected scores will be used."
#> [1] "Not correcting with APC as there are not enough interactions to estimate the background distribution. Uncorrected scores will be used."
#> [1] "Not correcting with APC as there are not enough interactions to estimate the background distribution. Uncorrected scores will be used."
#>    user  system elapsed 
#>   0.167   0.000   0.166 
#> [1] "Memory usage: 63.1 MB"
#> [1] "-> Performing wMI FDR analysis..."
#>    user  system elapsed 
#>   3.398   0.093   3.491 
#> [1] "Memory usage: 63.3 MB"
#> [1] "-> Calculating APC threshold..."

Intrepreting the results

  • Lets look into the results
  • The below table explains what each coloum means.
Colnames Meaning
SFE_1 Selected Functional Event (SFE_1)
SFE_2 Selected Functional Event (SFE_2)
name Interaction Motif
type_1 Type of mutation of SFE_1
type_2 Type of mutation of SFE_2
int_type Interaction Motif type
support_1 Samples mutated with SFE_1
support_2 Samples mutated with SFE_2
freq_1 Frequency of SFE_1
freq_2 Frequency of SFE_2
overlap Co-mutation between SFE_1 and SFE_2
max_overlap Maximum possible Co-mutation
freq_overlap Frequency of Co-mutation
r_overlap Background Co-mutation
r_freq_overlap Background frequency of Co-mutation
diff_overlap Difference of co-mutations
abs_diff_overlap Absoulte Difference of co-mutations
direction Interaction Type
wMI_stat Weighted Mutual Information
wMI_p.value P-value on Mutual Information
ME_p.value P-value on co-mutation
E.r.wMI_stat Background Weighted Mutual Information
MI_diff Difference of Mutual Information
wMI_p.value_FDR FDR
select_score_good_cancer_cell_2017_criterion_1 Cancer Cell Paper Criteria
select_score Effect Size (select Score) 
options(width=100)
# Look into dataframe
(alpi %>% filter(wMI_p.value_FDR) %>% arrange(desc(select_score))) %>% head(2) 
#>                                                                                                                  SFE_1
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493    AMP.consensus.chr14:35870717-36159897
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876 DEL.consensus.chr4:183089197-186421724
#>                                                                                                                  SFE_2
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493    AMP.consensus.chr14:37858832-38371493
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876 DEL.consensus.chr4:187186290-187647876
#>                                                                                                                                                            name
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493     AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876 DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876
#>                                                                                 type_1 type_2
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493      AMP    AMP
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876    DEL    DEL
#>                                                                                  int_type support_1
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493   AMP - AMP        57
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876 DEL - DEL        12
#>                                                                                 support_2
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493          53
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876        12
#>                                                                                     freq_1
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493   0.11656442
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876 0.02453988
#>                                                                                     freq_2 overlap
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493   0.10838446      47
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876 0.02453988      11
#>                                                                                 max_overlap
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493            53
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876          12
#>                                                                                 freq_overlap
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493      0.8867925
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876    0.9166667
#>                                                                                 r_overlap
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493       10.04
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876      0.72
#>                                                                                 r_freq_overlap
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493         0.189434
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876       0.060000
#>                                                                                 diff_overlap
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493          36.96
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876        10.28
#>                                                                                 abs_diff_overlap
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493              36.96
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876            10.28
#>                                                                                 direction  wMI_stat
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493          CO 0.4241418
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876        CO 0.3612067
#>                                                                                 wMI_p.value
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493             0
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876           0
#>                                                                                 ME_p.value
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493            0
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876          0
#>                                                                                 E.r.wMI_stat
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493    0.003527209
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876  0.016350097
#>                                                                                   MI_diff
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493   0.4206146
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876 0.3448566
#>                                                                                 wMI_p.value_FDR
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493              TRUE
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876            TRUE
#>                                                                                 select_score_good_cancer_cell_2017_criterion_1
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493                                             TRUE
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876                                           TRUE
#>                                                                                 select_score
#> AMP.consensus.chr14:35870717-36159897 - AMP.consensus.chr14:37858832-38371493      0.4005788
#> DEL.consensus.chr4:183089197-186421724 - DEL.consensus.chr4:187186290-187647876    0.3413985
# Total significant Hits
alpi  %>% filter(wMI_p.value_FDR) %>% count(wMI_p.value_FDR,direction)
#>   wMI_p.value_FDR direction   n
#> 1            TRUE        CO 108
#> 2            TRUE        ME  18

Suggestion

  • Please go through the methods on best way to filter and use the method as mentioned in the paper.
    • Mina, M., Iyer, A., Tavernari, D., Raynaud, F., & Ciriello, G. (2020). Discovering functional evolutionary dependencies in human cancers. Nature Genetics, 52(11), 1198-1207.

SessionInfo 

# Print the sessionInfo
sessionInfo()
#> R version 4.1.3 (2022-03-10)
#> Platform: x86_64-conda-linux-gnu (64-bit)
#> Running under: CentOS Linux 7 (Core)
#> 
#> Matrix products: default
#> BLAS/LAPACK: /home/arvind/miniconda3/envs/R_4/lib/libopenblasp-r0.3.18.so
#> 
#> locale:
#>  [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C               LC_TIME=en_US.UTF-8       
#>  [4] LC_COLLATE=en_US.UTF-8     LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
#>  [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                  LC_ADDRESS=C              
#> [10] LC_TELEPHONE=C             LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       
#> 
#> attached base packages:
#> [1] stats     graphics  grDevices utils     datasets  methods   base     
#> 
#> other attached packages:
#> [1] dplyr_1.1.1  select_1.6.3
#> 
#> loaded via a namespace (and not attached):
#>  [1] pillar_1.9.0       bslib_0.3.1        compiler_4.1.3     jquerylib_0.1.4    iterators_1.0.14  
#>  [6] tools_4.1.3        digest_0.6.29      tibble_3.2.1       jsonlite_1.7.2     evaluate_0.14     
#> [11] memoise_2.0.1      lifecycle_1.0.3    lattice_0.20-45    pkgconfig_2.0.3    rlang_1.1.0       
#> [16] Matrix_1.4-0       foreach_1.5.2      igraph_1.2.10      cli_3.6.1          parallel_4.1.3    
#> [21] yaml_2.2.1         pkgdown_2.0.7      xfun_0.29          fastmap_1.1.0      stringr_1.4.0     
#> [26] knitr_1.35         generics_0.1.2     desc_1.4.0         fs_1.5.2           vctrs_0.6.1       
#> [31] sass_0.4.0         systemfonts_1.0.4  tidyselect_1.2.0   rprojroot_2.0.2    grid_4.1.3        
#> [36] glue_1.6.2         BiRewire_3.26.0    R6_2.5.1           textshaping_0.3.6  fansi_1.0.3       
#> [41] rmarkdown_2.11     purrr_1.0.1        magrittr_2.0.3     matrixStats_0.61.0 codetools_0.2-18  
#> [46] htmltools_0.5.2    tsne_0.1-3         ragg_1.2.2         utf8_1.2.2         stringi_1.7.6     
#> [51] doParallel_1.0.17  cachem_1.0.6       slam_0.1-49        crayon_1.5.1