In CoGAPS/CoGAPS: Coordinated Gene Activity in Pattern Sets

library(CoGAPS)
library(BiocParallel)

Vignette Version

This vignette was built using CoGAPS version:

packageVersion("CoGAPS")

Introduction

Coordinated Gene Association in Pattern Sets (CoGAPS) is a technique for latent space learning in gene expression data. CoGAPS is a member of the Nonnegative Matrix Factorization (NMF) class of algorithms. NMFs factorize a data matrix into two related matrices containing gene weights, the Amplitude (A) matrix, and sample weights, the Pattern (P) Matrix. Each column of A or row of P defines a feature and together this set of features defines the latent space among genes and samples, respectively. In NMF, the values of the elements in the A and P matrices are constrained to be greater than or equal to zero. This constraint simultaneously reflects the non-negative nature of gene expression data and enforces the additive nature of the resulting feature dimensions, generating solutions that are biologically intuitive to interpret (@SEUNG_1999).

Software Setup

CoGAPS can be installed directly from the FertigLab Github Repository using R devtools or from Bioconductor

devtools::install_github("FertigLab/CoGAPS")

# To install via BioConductor:
install.packages("BiocManager")
BiocManager::install("FertigLab/CoGAPS")

When CoGAPS has installed correctly, you will see this message:

** installing vignettes

** testing if installed package can be loaded from temporary location

** checking absolute paths in shared objects and dynamic libraries

** testing if installed package can be loaded from final location

** testing if installed package keeps a record of temporary installation path

* DONE (CoGAPS)

Running CoGAPS on Simulated Toy Data

We first give a walkthrough of the package features using a simple, simulated data set. In later sections we provide two example workflows on real data sets.

Import the CoGAPS library with the following command:

library(CoGAPS)

To ensure CoGAPS is working properly, we will first load in the simulated toy data for a test run.

Single-cell data will be loaded later in this file.

data('modsimdata')
# input to CoGAPS
modsimdata

Next, we will set the parameters to be used by CoGAPS. First, we will create a CogapsParams object, then set parameters with the setParam function.

# create new parameters object
params <- new("CogapsParams")

# view all parameters
params

# get the value for a specific parameter
getParam(params, "nPatterns")

# set the value for a specific parameter
params <- setParam(params, "nPatterns", 3)
getParam(params, "nPatterns")

Run CoGAPS on the ModSim data.

Since this is a small dataset, the expected runtime is only about 5-10 seconds.

The only required argument to CoGAPS is the data set. This can be a matrix, data.frame, SummarizedExperiment, SingleCellExperiment or the path of a file (tsv, csv, mtx, gct) containing the data.

# run CoGAPS with specified parameters
cogapsresult <- CoGAPS(modsimdata, params, outputFrequency = 10000)

Verify that a similar output appears: ```{eval=FALSE} This is CoGAPS version 3.19.1 

Running Standard CoGAPS on modsimdata (25 genes and 20 samples) with parameters:

-- Standard Parameters -- nPatterns            3  nIterations          50000  seed                 622  sparseOptimization   FALSE 

-- Sparsity Parameters -- alpha          0.01  maxGibbsMass   100  Data Model: Dense, Normal Sampler Type: Sequential Loading Data...Done! (00:00:00)

-- Equilibration Phase -- 10000 of 50000, Atoms: 59(A), 49(P), ChiSq: 245, Time: 00:00:00 00:00:00 20000 of 50000, Atoms: 68(A), 46(P), ChiSq: 188, Time: 00:00:00 00:00:00 30000 of 50000, Atoms: 80(A), 47(P), ChiSq: 134, Time: 00:00:00 00:00:00 40000 of 50000, Atoms: 69(A), 46(P), ChiSq: 101, Time: 00:00:00 00:00:00 50000 of 50000, Atoms: 76(A), 53(P), ChiSq: 132, Time: 00:00:00 00:00:00

-- Sampling Phase --

10000 of 50000, Atoms: 82(A), 52(P), ChiSq: 94, Time: 00:00:00 00:00:00 20000 of 50000, Atoms: 74(A), 54(P), ChiSq: 144, Time: 00:00:01 00:00:01 30000 of 50000, Atoms: 79(A), 47(P), ChiSq: 116, Time: 00:00:01 00:00:01 40000 of 50000, Atoms: 79(A), 46(P), ChiSq: 132, Time: 00:00:01 00:00:01 50000 of 50000, Atoms: 76(A), 48(P), ChiSq: 124, Time: 00:00:01 00:00:01

This means that the underlying C++ library has run correctly, and
everything is installed how it should be.

We now examine the result object.

## Analyzing the Toy Data CoGAPS result

CoGAPS returns a object of the class `CogapsResult` which inherits from
`LinearEmbeddingMatrix` (defined in the `SingleCellExperiment` package).
CoGAPS stores the lower dimensional representation of the samples (P
matrix) in the `sampleFactors` slot and the weight of the features (A
matrix) in the `featureLoadings` slot. `CogapsResult` also adds two of
its own slots - `factorStdDev` and `loadingStdDev` which contain the
standard deviation across sample points for each matrix.

There is also some information in the `metadata` slot such as the
original parameters and value for the Chi-Sq statistic. In general, the
metadata will vary depending on how `CoGAPS` was called in the first
place. The package provides these functions for querying the metadata in
a safe manner:

```r
cogapsresult
cogapsresult@sampleFactors
cogapsresult@featureLoadings

# check reference result:
data('modsimresult')

If both matrices--sampleFactors and featureLoadings--have reasonable values (small, nonnegative, somewhat random-seeming), it is an indication that CoGAPS is working as expected.

We now continue with single-cell analysis.

Single-cell CoGAPS

# OPTION: download data object from Zenodo
options(timeout=50000) # adjust this if you're getting timeout downloading the file
bfc <- BiocFileCache::BiocFileCache()
pdac_url <- "https://zenodo.org/record/7709664/files/inputdata.Rds"
pdac_data <- BiocFileCache::bfcrpath(bfc, pdac_url)
pdac_data <- readRDS(pdac_data)
pdac_data

We also want to extract the counts matrix to provide directly to CoGAPS

pdac_epi_counts <- as.matrix(pdac_data@assays$originalexp@counts)
norm_pdac_epi_counts <- log1p(pdac_epi_counts)

head(pdac_epi_counts, n=c(5L, 2L))
head(norm_pdac_epi_counts, n=c(5L, 2L))

Most of the time we will set some parameters before running CoGAPS. Parameters are managed with a CogapsParams object. This object will store all parameters needed to run CoGAPS and provides a simple interface for viewing and setting the parameter values.

library(CoGAPS)
pdac_params <- CogapsParams(nIterations=50000, # 50000 iterations
                seed=42, # for consistency across stochastic runs
                nPatterns=8, # each thread will learn 8 patterns
                sparseOptimization=TRUE, # optimize for sparse data
                distributed="genome-wide") # parallelize across sets

pdac_params

If you wish to run distributed CoGAPS, which is recommended to improve the computational efficiency for most large datasets, you must also call the setDistributedParams function. For a complete description of the parallelization strategy used in distributed CoGAPS, please refer to our preprint: https://www.biorxiv.org/content/10.1101/2022.07.09.499398v1

pdac_params <- setDistributedParams(pdac_params, nSets=7)
pdac_params

With all parameters set, we are now ready to run CoGAPS. Please note that this is the most time-consuming step of the procedure. Timing can take several hours and scales nlog(n) based on dataset size, as well as the parameter values set for 'nPatterns' and 'nIterations'. Time is increased when learning more patterns, when running more iterations, and when running a larger dataset, with iterations having the largest variable impact on the runtime of the NMF function.

startTime <- Sys.time()

pdac_epi_result <- CoGAPS(pdac_epi_counts, pdac_params)
endTime <- Sys.time()

saveRDS(pdac_epi_result, "../data/pdac_epi_cogaps_result.Rds")

# To save as a .csv file, use the following line:
toCSV(pdac_epi_result, "../data")

Analyzing the CoGAPS result

Now that the CoGAPS run is complete, learned patterns can be investigated. Due to the stochastic nature of the MCMC sampling in CoGAPS and long run time, it is generally a good idea to immediately save your CoGAPS result object to a file to have (Box 17), then read it in for downstream analysis.

If you wish to load and examine a precomputed result object, please do so by:

# OPTION: download precomputed CoGAPS result object from Zenodo
#caches download of the hosted file
cogapsresult_url <- "https://zenodo.org/record/7709664/files/cogapsresult.Rds"
cogapsresult <- BiocFileCache::bfcrpath(bfc, cogapsresult_url)
cogapsresult <- readRDS(cogapsresult)

To load your own result, simply edit the file path:

library(CoGAPS)
cogapsresult <- readRDS("../data/pdac_epi_cogaps_result.Rds")

It is recommended to immediately visualize pattern weights on a UMAP because you will immediately see whether they are showing strong signal and make common sense.

Since pattern weights are all continuous and nonnegative, they can be used to color a UMAP in the same way as one would color by gene expression. The sampleFactors matrix is essentially just nPatterns different annotations for each cell, and featureLoadings is likewise just nPatterns annotations for each gene. This makes it very simple to incorporate pattern data into any data structure and workflow.

Load CoGAPS pattern information into Seurat object

To store CoGAPS patterns as an Assay within a Seurat object (recommended):

library(Seurat)
# make sure pattern matrix is in same order as the input data
patterns_in_order <-t(cogapsresult@sampleFactors[colnames(pdac_data),])

# add CoGAPS patterns as an assay
pdac_data[["CoGAPS"]] <- CreateAssayObject(counts = patterns_in_order)

Plot patterns on an embedding

With the help of Seurat's FeaturePlot function, we generate a UMAP embedding of the cells colored by the intensity of each pattern.

DefaultAssay(pdac_data) <- "CoGAPS"
pattern_names = rownames(pdac_data@assays$CoGAPS)

library(viridis)
color_palette <- viridis(n=10)

FeaturePlot(pdac_data, pattern_names, cols=color_palette, reduction = "umap") & NoLegend()

Pattern Markers

We provide a patternMarkers() CoGAPS function to find markers (i.e. genes or samples) most associated with each pattern. It returns a per-pattern dictionary of markers, their ranking, and their distance "score" for each pattern.

Marker assignment methods

patternMarkers() can run in two marker selection modes, controlled with the "threshold" parameter, which can be set to "all" or "cut". In both cases the ranking of markers is essentially sorting Euclidian distances between a candidate marker and a synthetic one-row matrix that represents an "ideal" pattern.

If threshold="all" (default), each gene is treated as a marker of exactly one pattern, whichever it is most strongly associated with, based on the distance metric.

For the threshold="cut", the candidates are sorted by best intra-pattern rank and are added to markers list unitil a certain rank is reached. This rank corresponds to the first occurence of intra-pattern rank being worse than the inter-pattern rank. This mode usually yields much less markers per pattern, and the same marker can appear in multiple patterns.

Axis selection

The axis parameter can be used to select the axis along which the markers are selected. By default, axis=1, which means that the feature (i.e. gene) markers are estimated. If axis=2, the sample markers are estimated.

The three components of the returned dictionary pm are:

patternMarkers() Output

Output is a list of three components:

PatternMarkers:

• A list of marker genes for each pattern

PatternRanks:

• Each marker gene ranked by association for each pattern
• Whole natural numbers, assigning each marker a place in the rank for each pattern.
• Lower rank indicates higher association and vice versa.

PatternScores:

• Scores describe how strongly a marker gene is associated with a pattern.
• A higher score value indicates the marker gene is less associated with the pattern (as score is a distance) metric, and vice versa.
• Scores have nonnegative values mostly falling between 0 and 2

Example

pm <- patternMarkers(cogapsresult)

Pattern GSEA

One way to explore and use CoGAPS patterns is to conduct gene set enrichment analysis by functionally annotating the genes which are significant for each pattern. The getPatternGeneSet function provides a wrapper around the gsea and fora methods from fgsea for gene set enrichment in CoGAPS pattern amplitudes or gene set overrepresentation in pattern markers, respectively. Gene sets for testing are provided as a list to allow testing of gene sets from many sources, such as hallmark gene sets from msigDB or gene ontology sets from org.Hs.eg.db.

To perform gene set overrepresentation on pattern markers, please run:

hallmark_ls <- list(
  "HALLMARK_ALLOGRAFT_REJECTION" = c(
    "AARS1","ABCE1","ABI1","ACHE","ACVR2A","AKT1","APBB1","B2M","BCAT1","BCL10","BCL3","BRCA1","C2","CAPG","CARTPT","CCL11","CCL13","CCL19","CCL2","CCL22","CCL4","CCL5","CCL7","CCND2","CCND3","CCR1","CCR2","CCR5","CD1D","CD2","CD247","CD28","CD3D","CD3E","CD3G","CD4","CD40","CD40LG","CD47","CD7","CD74","CD79A","CD80","CD86","CD8A","CD8B","CD96","CDKN2A","CFP","CRTAM","CSF1","CSK","CTSS","CXCL13","CXCL9","CXCR3","DARS1","DEGS1","DYRK3","EGFR","EIF3A","EIF3D","EIF3J","EIF4G3","EIF5A","ELANE","ELF4","EREG","ETS1","F2","F2R","FAS","FASLG","FCGR2B","FGR","FLNA","FYB1","GALNT1","GBP2","GCNT1","GLMN","GPR65","GZMA","GZMB","HCLS1","HDAC9","HIF1A","HLA-A","HLA-DMA","HLA-DMB","HLA-DOA","HLA-DOB","HLA-DQA1","HLA-DRA","HLA-E","HLA-G","ICAM1","ICOSLG","IFNAR2","IFNG","IFNGR1","IFNGR2","IGSF6","IKBKB","IL10","IL11","IL12A","IL12B","IL12RB1","IL13","IL15","IL16","IL18","IL18RAP","IL1B","IL2","IL27RA","IL2RA","IL2RB","IL2RG","IL4","IL4R","IL6","IL7","IL9","INHBA","INHBB","IRF4","IRF7","IRF8","ITGAL","ITGB2","ITK","JAK2","KLRD1","KRT1","LCK","LCP2","LIF","LTB","LY75","LY86","LYN","MAP3K7","MAP4K1","MBL2","MMP9","MRPL3","MTIF2","NCF4","NCK1","NCR1","NLRP3","NME1","NOS2","NPM1","PF4","PRF1","PRKCB","PRKCG","PSMB10","PTPN6","PTPRC","RARS1","RIPK2","RPL39","RPL3L","RPL9","RPS19","RPS3A","RPS9","SIT1","SOCS1","SOCS5","SPI1","SRGN","ST8SIA4","STAB1","STAT1","STAT4","TAP1","TAP2","TAPBP","TGFB1","TGFB2","THY1","TIMP1","TLR1","TLR2","TLR3","TLR6","TNF","TPD52","TRAF2","TRAT1","UBE2D1","UBE2N","WARS1","WAS","ZAP70"
    ),
  "HALLMARK_EPITHELIAL_MESENCHYMAL_TRANSITION" = c(
    "ABI3BP","ACTA2","ADAM12","ANPEP","APLP1","AREG","BASP1","BDNF","BGN","BMP1","CADM1","CALD1","CALU","CAP2","CAPG","CD44","CD59","CDH11","CDH2","CDH6","COL11A1","COL12A1","COL16A1","COL1A1","COL1A2","COL3A1","COL4A1","COL4A2","COL5A1","COL5A2","COL5A3","COL6A2","COL6A3","COL7A1","COL8A2","COMP","COPA","CRLF1","CCN2","CTHRC1","CXCL1","CXCL12","CXCL6","CCN1","DAB2","DCN","DKK1","DPYSL3","DST","ECM1","ECM2","EDIL3","EFEMP2","ELN","EMP3","ENO2","FAP","FAS","FBLN1","FBLN2","FBLN5","FBN1","FBN2","FERMT2","FGF2","FLNA","FMOD","FN1","FOXC2","FSTL1","FSTL3","FUCA1","FZD8","GADD45A","GADD45B","GAS1","GEM","GJA1","GLIPR1","COLGALT1","GPC1","GPX7","GREM1","HTRA1","ID2","IGFBP2","IGFBP3","IGFBP4","IL15","IL32","IL6","CXCL8","INHBA","ITGA2","ITGA5","ITGAV","ITGB1","ITGB3","ITGB5","JUN","LAMA1","LAMA2","LAMA3","LAMC1","LAMC2","P3H1","LGALS1","LOX","LOXL1","LOXL2","LRP1","LRRC15","LUM","MAGEE1","MATN2","MATN3","MCM7","MEST","MFAP5","MGP","MMP1","MMP14","MMP2","MMP3","MSX1","MXRA5","MYL9","MYLK","NID2","NNMT","NOTCH2","NT5E","NTM","OXTR","PCOLCE","PCOLCE2","PDGFRB","PDLIM4","PFN2","PLAUR","PLOD1","PLOD2","PLOD3","PMEPA1","PMP22","POSTN","PPIB","PRRX1","PRSS2","PTHLH","PTX3","PVR","QSOX1","RGS4","RHOB","SAT1","SCG2","SDC1","SDC4","SERPINE1","SERPINE2","SERPINH1","SFRP1","SFRP4","SGCB","SGCD","SGCG","SLC6A8","SLIT2","SLIT3","SNAI2","SNTB1","SPARC","SPOCK1","SPP1","TAGLN","TFPI2","TGFB1","TGFBI","TGFBR3","TGM2","THBS1","THBS2","THY1","TIMP1","TIMP3","TNC","TNFAIP3","TNFRSF11B","TNFRSF12A","TPM1","TPM2","TPM4","VCAM1","VCAN","VEGFA","VEGFC","VIM","WIPF1","WNT5A"
  )
)

hallmarks_ora <- getPatternGeneSet(cogapsresult,
                                   gene.sets = hallmark_ls,
                                   method = "overrepresentation")

hallmarks is a list of data frames, each containing hallmark overrepresentation statistics corresponding to one pattern.

To generate a barchart of the most significant hallmarks for any given pattern, please run:

pl_pattern7 <- plotPatternGeneSet(
  patterngeneset = hallmarks_ora, whichpattern = 7, padj_threshold = 0.05
)
pl_pattern7

To generate statistics on the association between certain sample groups and patterns, we provide a wrapper function, called MANOVA.This will allow us to explore if the patterns we have discovered lend to statistically significant differences in the sample groups. We will first load in the original data (if not already done earlier).

Then, create a new matrix called “interestedVariables” consisting of the metadata variables of interest in conducting analysis on. Lastly, call the wrapper function, passing in the result object as well.

# create dataframe of interested variables 
interestedVariables <- cbind([e-mail][["nCount_RNA"]], [e-mail][["nFeature_RNA"]])
# call cogaps manova wrapper
manova_results <- MANOVA(interestedVariables, cogapsresult)

The function will print out the MANOVA results for each pattern learned based on the variables of interest. From the output, we can observe that all p-values have a value of 0.0, indicating that differences observed in the sample groups based on the patterns are statistically significant.

Citing CoGAPS

If you use the CoGAPS package for your analysis, please cite @FERTIG_2010

If you use the gene set statistic, please cite @OCHS_2009

References

CoGAPS/CoGAPS documentation built on June 23, 2024, 5:19 p.m.