vignettes/MLdatasets.Rmd
MLdatasets.RmdThis vignette identifies and writes to disk datasets for use in multi-dataset machine learning. For each relevant study condition, two files are written:
.rda file containing a TreeSummarizedExperiment with
full colData, rowData, a phylogenetic tree, and assay data containing
taxonomic data..csv.gz file containing only key sample metadata, and
taxonomic data (variables in columns)Do
BiocManager::install("waldronlab/curatedMetagenomicAnalyses")
to install the makeSEforCondition function and to run this
vignette.
Packages used here:
These are the 10 study conditions most commonly found in curatedMetagenomicData:
data("sampleMetadata")
availablediseases <- pull(sampleMetadata, study_condition) %>%
table() %>%
sort(decreasing = TRUE)
availablediseases## .
## control IBD CRC
## 14446 2086 702
## premature_born T2D IGT
## 448 332 288
## ACVD adenoma cirrhosis
## 214 209 132
## STH otitis schizofrenia
## 108 107 106
## hypertension AS T1D
## 99 97 89
## melanoma acute_diarrhoea pre-hypertension
## 64 56 56
## ME/CFS migraine STEC
## 50 49 42
## fatty_liver psoriasis carcinoma_surgery_history
## 41 41 40
## FMT AD cephalosporins
## 40 38 36
## CDI asthma periodontitis
## 33 24 24
## SRP peri-implantitis BD
## 24 23 20
## mucositis bronchitis respiratoryinf
## 20 18 13
## metabolic_syndrome pyelonephritis infectiousgastroenteritis
## 10 6 5
## fever pneumonia tonsillitis
## 3 3 3
## cough pyelonefritis skininf
## 2 2 2
## stomatitis cystitis NK
## 2 1 1
## salmonellosis sepsis suspinf
## 1 1 1And the number of studies they are found in:
studies <- lapply(names(availablediseases), function(x){
filter(sampleMetadata, study_condition %in% x) %>%
pull(study_name) %>%
unique()
})
names(studies) <- names(availablediseases)
studies <- studies[-grep("control", names(studies))] #get rid of controls
studies <- studies[sapply(studies, length) > 1] #available in more than one study
studies## $IBD
## [1] "HallAB_2017" "HMP_2019_ibdmdb" "IjazUZ_2017" "LiJ_2014"
## [5] "NielsenHB_2014" "VilaAV_2018"
##
## $CRC
## [1] "FengQ_2015" "GuptaA_2019" "HanniganGD_2017" "ThomasAM_2018a"
## [5] "ThomasAM_2018b" "ThomasAM_2019_c" "VogtmannE_2016" "WirbelJ_2018"
## [9] "YachidaS_2019" "YuJ_2015" "ZellerG_2014"
##
## $premature_born
## [1] "BrooksB_2017" "OlmMR_2017"
##
## $T2D
## [1] "HMP_2019_t2d" "KarlssonFH_2013" "LiJ_2014"
## [4] "QinJ_2012" "SankaranarayananK_2015"
##
## $IGT
## [1] "HMP_2019_t2d" "KarlssonFH_2013"
##
## $adenoma
## [1] "FengQ_2015" "HanniganGD_2017" "ThomasAM_2018a" "YachidaS_2019"
## [5] "ZellerG_2014"
##
## $cirrhosis
## [1] "LoombaR_2017" "QinN_2014"
##
## $STH
## [1] "RosaBA_2018" "RubelMA_2020"
##
## $schizofrenia
## [1] "Castro-NallarE_2015" "ZhuF_2020"
##
## $T1D
## [1] "Heitz-BuschartA_2016" "KosticAD_2015" "LiJ_2014"
##
## $melanoma
## [1] "GopalakrishnanV_2018" "MatsonV_2018"
##
## $acute_diarrhoea
## [1] "DavidLA_2015" "KieserS_2018"Each of these datasets has six data types associated with it; for example:
curatedMetagenomicData("JieZ_2017.+")## 2021-03-31.JieZ_2017.gene_families
## 2021-03-31.JieZ_2017.marker_abundance
## 2021-03-31.JieZ_2017.marker_presence
## 2021-03-31.JieZ_2017.pathway_abundance
## 2021-03-31.JieZ_2017.pathway_coverage
## 2021-03-31.JieZ_2017.relative_abundance
## 2021-10-14.JieZ_2017.gene_families
## 2021-10-14.JieZ_2017.marker_abundance
## 2021-10-14.JieZ_2017.marker_presence
## 2021-10-14.JieZ_2017.pathway_abundance
## 2021-10-14.JieZ_2017.pathway_coverage
## 2021-10-14.JieZ_2017.relative_abundance
for (i in seq_along(studies)){
cond <- names(studies)[i]
se <-
curatedMetagenomicAnalyses::makeSEforCondition(cond, removestudies = "HMP_2019_ibdmdb", dataType = "relative_abundance")
print(paste("Next study condition:", cond, " /// Body site: ", unique(colData(se)$body_site)))
print(with(colData(se), table(study_name, study_condition)))
cat("\n \n")
save(se, file = paste0(cond, ".rda"))
flattext <- select(as.data.frame(colData(se)), c("study_name", "study_condition", "subject_id"))
rownames(flattext) <- colData(se)$sample_id
flattext <- cbind(flattext, data.frame(t(assay(se))))
write.csv(flattext, file = paste0(cond, ".csv"))
system(paste0("gzip ", cond, ".csv"))
}## [1] "Next study condition: IBD /// Body site: stool"
## study_condition
## study_name control IBD
## HallAB_2017 74 185
## IjazUZ_2017 38 56
## LiJ_2014 10 140
## NielsenHB_2014 248 148
## VilaAV_2018 0 355
##
##
## [1] "Next study condition: CRC /// Body site: stool"
## study_condition
## study_name control CRC
## FengQ_2015 61 46
## GuptaA_2019 30 30
## HanniganGD_2017 28 27
## ThomasAM_2018a 24 29
## ThomasAM_2018b 28 32
## ThomasAM_2019_c 40 40
## VogtmannE_2016 52 52
## WirbelJ_2018 65 60
## YachidaS_2019 251 258
## YuJ_2015 53 75
## ZellerG_2014 61 53
##
##
## [1] "Next study condition: premature_born /// Body site: stool"
## [2] "Next study condition: premature_born /// Body site: oralcavity"
## [3] "Next study condition: premature_born /// Body site: skin"
## study_condition
## study_name control premature_born
## BrooksB_2017 5 403
## OlmMR_2017 0 45
##
##
## [1] "Next study condition: T2D /// Body site: stool"
## study_condition
## study_name control T2D
## HMP_2019_t2d 46 11
## KarlssonFH_2013 43 53
## LiJ_2014 10 79
## QinJ_2012 174 170
## SankaranarayananK_2015 18 19
##
##
## [1] "Next study condition: IGT /// Body site: stool"
## study_condition
## study_name control IGT
## HMP_2019_t2d 46 239
## KarlssonFH_2013 43 49
##
##
## [1] "Next study condition: adenoma /// Body site: stool"
## study_condition
## study_name adenoma control
## FengQ_2015 47 61
## HanniganGD_2017 26 28
## ThomasAM_2018a 27 24
## YachidaS_2019 67 251
## ZellerG_2014 42 61
##
##
## [1] "Next study condition: cirrhosis /// Body site: stool"
## study_condition
## study_name cirrhosis control
## LoombaR_2017 9 36
## QinN_2014 123 114
##
##
## [1] "Next study condition: STH /// Body site: stool"
## study_condition
## study_name control STH
## RosaBA_2018 5 19
## RubelMA_2020 86 89
##
##
## [1] "Next study condition: schizofrenia /// Body site: oralcavity"
## [2] "Next study condition: schizofrenia /// Body site: stool"
## study_condition
## study_name control schizofrenia
## Castro-NallarE_2015 16 16
## ZhuF_2020 81 90
##
##
## [1] "Next study condition: T1D /// Body site: stool"
## study_condition
## study_name control T1D
## Heitz-BuschartA_2016 26 27
## KosticAD_2015 89 31
## LiJ_2014 10 31
##
##
## [1] "Next study condition: melanoma /// Body site: stool"
## study_condition
## study_name melanoma
## GopalakrishnanV_2018 25
## MatsonV_2018 39
##
##
## [1] "Next study condition: acute_diarrhoea /// Body site: stool"
## study_condition
## study_name acute_diarrhoea control
## DavidLA_2015 38 9
## KieserS_2018 18 9
##
## Download the .csv and .rda files directly from https://www.dropbox.com/sh/0t0nbhj9eqm3wkq/AACZIw42WA-uHjzo97bG5tE6a?dl=0