TMRC3 Comprehensive Data Analysis: 202106

samplesheet <- "sample_sheets/tmrc3_samples_20210610.xlsx"

1 Annotation

We take the annotation data from ensembl’s biomart instance. The genome which was used to map the data was hg38 revision 100. My default when using biomart is to load the data from 1 year before the current date.

hs_annot <- sm(load_biomart_annotations(year = "2020"))
hs_annot <- hs_annot[["annotation"]]
hs_annot[["transcript"]] <- paste0(rownames(hs_annot), ".", hs_annot[["version"]])
rownames(hs_annot) <- make.names(hs_annot[["ensembl_gene_id"]], unique = TRUE)
tx_gene_map <- hs_annot[, c("transcript", "ensembl_gene_id")]

summary(hs_annot)

##  ensembl_transcript_id ensembl_gene_id       version     transcript_version
##  Length:227921         Length:227921      Min.   : 1.0   Min.   : 1.00     
##  Class :character      Class :character   1st Qu.: 6.0   1st Qu.: 1.00     
##  Mode  :character      Mode  :character   Median :12.0   Median : 1.00     
##                                           Mean   :10.7   Mean   : 3.08     
##                                           3rd Qu.:16.0   3rd Qu.: 5.00     
##                                           Max.   :29.0   Max.   :17.00     
##                                                                            
##  hgnc_symbol        description        gene_biotype         cds_length    
##  Length:227921      Length:227921      Length:227921      Min.   :     3  
##  Class :character   Class :character   Class :character   1st Qu.:   357  
##  Mode  :character   Mode  :character   Mode  :character   Median :   694  
##                                                           Mean   :  1139  
##                                                           3rd Qu.:  1446  
##                                                           Max.   :107976  
##                                                           NA's   :127343  
##  chromosome_name       strand          start_position      end_position     
##  Length:227921      Length:227921      Min.   :5.77e+02   Min.   :6.47e+02  
##  Class :character   Class :character   1st Qu.:3.11e+07   1st Qu.:3.12e+07  
##  Mode  :character   Mode  :character   Median :6.04e+07   Median :6.06e+07  
##                                        Mean   :7.41e+07   Mean   :7.42e+07  
##                                        3rd Qu.:1.09e+08   3rd Qu.:1.09e+08  
##                                        Max.   :2.49e+08   Max.   :2.49e+08  
##                                                                             
##   transcript       
##  Length:227921     
##  Class :character  
##  Mode  :character  
##                    
##                    
##                    
## 

hs_go <- sm(load_biomart_go()[["go"]])
hs_length <- hs_annot[, c("ensembl_gene_id", "cds_length")]
colnames(hs_length) <- c("ID", "length")

2 Introduction

This document is intended to provide an overview of TMRC3 samples which have been sequenced. It includes some plots and analyses showing the relationships among the samples as well as some differential analyses when possible.

3 Sample Estimation

3.1 Generate expressionsets

The sample sheet is copied from our shared online sheet and updated with each release of sequencing data.

3.1.1 Hisat2 expressionsets

The first thing to note is the large range in coverage. There are multiple samples with coverage which is too low to use. These will be removed shortly.

In the following block I immediately exclude any non-coding reads as well.

## Create the expressionset and immediately pass it to a filter
## removing the non protein coding genes.
sanitize_columns <- c("visitnumber", "clinicaloutcome", "donor",
                      "typeofcells", "clinicalpresentation",
                      "condition", "batch")
hs_expt <- create_expt(samplesheet,
                       file_column = "hg38100hisatfile",
                       savefile = glue::glue("rda/hs_expt_all-v{ver}.rda"),
                       gene_info = hs_annot) %>%
  exclude_genes_expt(column = "gene_biotype", method = "keep",
                     pattern = "protein_coding", meta_column = "ncrna_lost") %>%
  sanitize_expt_metadata(columns = sanitize_columns) %>%
  set_expt_factors(columns = sanitize_columns, class = "factor")

## Reading the sample metadata.

## Dropped 71 rows from the sample metadata because they were blank.

## The sample definitions comprises: 173 rows(samples) and 75 columns(metadata fields).

## Warning in create_expt(samplesheet, file_column = "hg38100hisatfile", savefile =
## glue::glue("rda/hs_expt_all-v{ver}.rda"), : Some samples were removed when cross
## referencing the samples against the count data.

## Matched 21452 annotations and counts.

## Bringing together the count matrix and gene information.

## Some annotations were lost in merging, setting them to 'undefined'.

## The final expressionset has 21481 rows and 155 columns.

## Before removal, there were 21481 genes, now there are 19941.

## There are 17 samples which kept less than 90 percent counts.

## TMRC30015 TMRC30017 TMRC30019 TMRC30044 TMRC30045 TMRC30154 TMRC30097 TMRC30075 
##     79.24     85.72     89.75     80.34     73.33     83.20     89.90     86.97 
## TMRC30087 TMRC30101 TMRC30104 TMRC30114 TMRC30127 TMRC30120 TMRC30128 TMRC30131 
##     83.63     88.41     80.29     87.62     89.49     79.16     82.53     86.82 
## TMRC30073 
##     89.26

levels(pData(hs_expt[["expressionset"]])[["visitnumber"]]) <- list(
    '0' = "notapplicable", '1' = 1, '2' = 2, '3' = 3)

Split this data into CDS and lncRNA. Oh crap in order to do that I need to recount the data. Running now (20210518)

## lnc_expt <- create_expt(samplesheet,
##                         file_column = "hg38100lncfile",
##                         gene_info = hs_annot)

3.1.1.1 Initial metrics

Once the data was loaded, there are a couple of metrics which may be plotted immediately.

nonzero <- plot_nonzero(hs_expt)
nonzero$plot

## Warning: ggrepel: 127 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps

ncrna_lost_df <- as.data.frame(pData(hs_expt)[["ncrna_lost"]])
rownames(ncrna_lost_df) <- rownames(pData(hs_expt))
colnames(ncrna_lost_df) <- "ncrna_lost"

tmpdf <- merge(nonzero$table, ncrna_lost_df, by = "row.names")
rownames(tmpdf) <- tmpdf[["Row.names"]]
tmpdf[["Row.names"]] <- NULL

ggplot(tmpdf, aes(x=ncrna_lost, y=nonzero_genes)) +
  ggplot2::geom_point() +
  ggplot2::ggtitle("Nonzero genes with respect to percent counts 
lost when ncRNA was removed.")

Najib doesn’t want this plot, but I am using it to check new samples, so will hide it from general use.

libsize <- plot_libsize(hs_expt)
libsize$plot

3.2 Minimum coverage sample filtering

I arbitrarily chose 11,000 non-zero genes as a minimum. We may want this to be higher.

hs_valid <- subset_expt(hs_expt, nonzero = 11000)

## The samples (and read coverage) removed when filtering 11000 non-zero genes are:

## TMRC30010 TMRC30050 TMRC30052 
##     52471    808149   3087347

## subset_expt(): There were 155, now there are 152 samples.

valid_write <- sm(write_expt(hs_valid, excel = glue("excel/hs_valid-v{ver}.xlsx")))

4 Project Aims

The project seeks to determine the relationship of the innate immune response and inflammatory signaling to the clinical outcome of antileishmanial drug treatment. We will test the hypothesis that the profile of innate immune cell activation and their dynamics through the course of treatment differ between CL patients with prospectively determined therapeutic cure or failure.

This will be achieved through the characterization of the in vivo dynamics of blood-derived monocyte, neutrophil and eosinophil transcriptome before, during and at the end of treatment in CL patients. Cell-type specific transcriptomes, composite signatures and time-response expression profiles will be contrasted among patients with therapeutic cure or failure.

4.1 Preparation

To address these, I added to the end of the sample sheet columns named ‘condition’, ‘batch’, ‘donor’, and ‘time’. These are filled in with shorthand values according to the above.

4.2 Global view

Before addressing the questions explicitly by subsetting the data, I want to get a look at the samples as they are.

new_names <- pData(hs_valid)[["samplename"]]
hs_valid <- hs_valid %>%
  set_expt_batches(fact = "cellssource") %>%
  set_expt_conditions(fact = "typeofcells") %>%
  set_expt_samplenames(newnames = new_names)

all_norm <- sm(normalize_expt(hs_valid, transform = "log2", norm = "quant",
                              convert = "cpm", filter = TRUE))

all_pca <- plot_pca(all_norm, plot_labels = FALSE,
                    plot_title = "PCA - Cell type", size_column = "visitnumber")
pp(file = glue("images/tmrc3_pca_nolabels-v{ver}.png"), image = all_pca$plot)

write.csv(all_pca$table, file = "coords/hs_donor_pca_coords.csv")
plot_corheat(all_norm, plot_title = "Heirarchical clustering:
         cell types")$plot

4.3 Examine samples relevant to clinical outcome

Now let us consider only the samples for which we have a clinical outcome. These fall primarily into either ‘cured’ or ‘failed’, but some people have not yet returned to the clinic after the first or second visit. These are deemed ‘lost’.

hs_clinical <- hs_valid %>%
  set_expt_conditions(fact = "clinicaloutcome") %>%
  set_expt_batches(fact = "typeofcells") %>%
  subset_expt(subset = "typeofcells!='pbmcs'&typeofcells!='macrophages'")

## subset_expt(): There were 152, now there are 132 samples.

chosen_colors <- c("#D95F02", "#7570B3", "#1B9E77", "#FF0000", "#FF0000")
names(chosen_colors) <- c("cure", "failure", "lost", "null", "notapplicable")
hs_clinical <- set_expt_colors(hs_clinical, colors = chosen_colors)

newnames <- make.names(pData(hs_clinical)[["samplename"]], unique = TRUE)
hs_clinical <- set_expt_samplenames(hs_clinical, newnames = newnames)

## Error in names(new_expt[["colors"]]) <- newnames: 'names' attribute [132] must be the same length as the vector [131]

hs_clinical_norm <- sm(normalize_expt(hs_clinical, filter = TRUE, transform = "log2",
                                      convert = "cpm", norm = "quant"))
clinical_pca <- plot_pca(hs_clinical_norm, plot_labels = FALSE,
                         size_column = "visitnumber", cis = NULL,
                         plot_title = "PCA - clinical samples")

## Error in data.frame(sampleid = as.character(design[["sampleid"]]), condition = design[[cond_column]], : arguments imply differing number of rows: 132, 131

pp(file = glue("images/all_clinical_nobatch_pca-v{ver}.png"), image = clinical_pca$plot,
   height = 8, width = 20)

## Error in pp(file = glue("images/all_clinical_nobatch_pca-v{ver}.png"), : object 'clinical_pca' not found

4.3.1 Repeat without the biopsy samples

hs_clinical_nobiop <- hs_clinical %>%
  subset_expt(subset = "typeofcells!='biopsy'")

## subset_expt(): There were 132, now there are 83 samples.

hs_clinical_nobiop_norm <- sm(normalize_expt(hs_clinical_nobiop, filter = TRUE, transform = "log2",
                                             convert = "cpm", norm = "quant"))
clinical_nobiop_pca <- plot_pca(hs_clinical_nobiop_norm, plot_labels = FALSE, cis = NULL,
                                plot_title = "PCA - clinical samples without biopsies")

## Potentially check over the experimental design, there appear to be missing values.

## Warning in plot_pca(hs_clinical_nobiop_norm, plot_labels = FALSE, cis = NULL, :
## There are NA values in the component data. This can lead to weird plotting
## errors.

pp(file = glue("images/all_clinical_nobiop_nobatch_pca-v{ver}.png"),
   image = clinical_nobiop_pca$plot)

4.3.2 Attempt to correct for the surrogate variables

At this time we have two primary data structures of interest: hs_clinical and hs_clinical_nobiop

hs_clinical_nb <- normalize_expt(hs_clinical, filter = TRUE, batch = "svaseq",
                                 transform = "log2", convert = "cpm")

## Removing 5271 low-count genes (14670 remaining).

## batch_counts: Before batch/surrogate estimation, 122582 entries are x==0: 6%.

## batch_counts: Before batch/surrogate estimation, 355614 entries are 0<x<1: 18%.

## Setting 23843 low elements to zero.

## transform_counts: Found 23843 values equal to 0, adding 1 to the matrix.

clinical_batch_pca <- plot_pca(hs_clinical_nb, plot_labels = FALSE, cis = NULL,
                               size_column = "visitnumber", plot_title = "PCA - clinical samples")

## Error in data.frame(sampleid = as.character(design[["sampleid"]]), condition = design[[cond_column]], : arguments imply differing number of rows: 132, 131

clinical_batch_pca$plot

## Error in eval(expr, envir, enclos): object 'clinical_batch_pca' not found

hs_clinical_nobiop_nb <- sm(normalize_expt(hs_clinical_nobiop, filter = TRUE, batch = "svaseq",
                                           transform = "log2", convert = "cpm"))
clinical_nobiop_batch_pca <- plot_pca(hs_clinical_nobiop_nb,
                                      plot_title = "PCA - clinical samples without biopsies",
                                      plot_labels = FALSE)

## Potentially check over the experimental design, there appear to be missing values.

## Warning in plot_pca(hs_clinical_nobiop_nb, plot_title = "PCA - clinical samples
## without biopsies", : There are NA values in the component data. This can lead to
## weird plotting errors.

pp(file = "images/clinical_batch.png", image = clinical_nobiop_batch_pca$plot)

test <- plot_pca(hs_clinical_nobiop_nb, size_column = "visitnumber",
                 plot_title = "PCA - clinical samples without biopsies",
                 plot_labels = FALSE)

## Potentially check over the experimental design, there appear to be missing values.

## Warning in plot_pca(hs_clinical_nobiop_nb, size_column = "visitnumber", : There
## are NA values in the component data. This can lead to weird plotting errors.

test$plot

clinical_nobiop_batch_tsne <- plot_tsne(hs_clinical_nobiop_nb,
                                        plot_title = "tSNE - clinical samples without biopsies",
                                        plot_labels = FALSE)

## Potentially check over the experimental design, there appear to be missing values.

## Warning in plot_pca(..., pc_method = "tsne"): There are NA values in the
## component data. This can lead to weird plotting errors.

clinical_nobiop_batch_tsne$plot

4.3.2.1 Look at remaining variance with variancePartition

test <- simple_varpart(hs_clinical_nobiop)

## 
## Total:111 s

test$partition_plot

4.4 Perform DE of the clinical samples cure vs. fail

individual_celltypes <- subset_expt(hs_clinical_nobiop, subset="condition!='lost'")

## subset_expt(): There were 83, now there are 68 samples.

hs_clinic_de <- sm(all_pairwise(individual_celltypes, model_batch = "svaseq", filter = TRUE))

hs_clinic_table <- sm(combine_de_tables(
    hs_clinic_de,
    excel = glue::glue("excel/individual_celltypes_table-v{ver}.xlsx")))

hs_clinic_sig <- sm(extract_significant_genes(
    hs_clinic_table,
    excel = glue::glue("excel/individual_celltypes_sig-v{ver}.xlsx")))

hs_clinic_sig[["summary_df"]]

##   limma_V1 limma_V2 edger_V1 edger_V2 deseq_V1 deseq_V2 ebseq_V1 ebseq_V2
## 1      183      181      263      213      249      232       96      166
##   basic_V1 basic_V2
## 1       46       26

hs_clinic_de[["comparison"]][["heat"]]

## NULL

4.4.1 Perform LRT with the clinical samples

I am not sure if we have enough samples across the three visit to completely work as well as we would like, but there is only 1 way to find out! Now that I think about it, one thing which might be awesome is to use cell type as an interacting factor…

4.4.1.1 With biopsy samples

I figure this might be a place where the biopsy samples might prove useful.

clinical_nolost <- subset_expt(hs_clinical, subset="condition!='lost'")

## subset_expt(): There were 132, now there are 115 samples.

lrt_visit_clinical_test <- deseq_lrt(clinical_nolost, transform = "vst",
                                     interactor_column = "visitnumber",
                                     interest_column = "clinicaloutcome")

## converting counts to integer mode

## Error in checkFullRank(modelMatrix): the model matrix is not full rank, so the model cannot be fit as specified.
##   Levels or combinations of levels without any samples have resulted in
##   column(s) of zeros in the model matrix.
## 
##   Please read the vignette section 'Model matrix not full rank':
## 
##   vignette('DESeq2')

lrt_visit_clinical_test[["favorite_genes"]]

## Error in eval(expr, envir, enclos): object 'lrt_visit_clinical_test' not found

lrt_celltype_clinical_test <- deseq_lrt(clinical_nolost, transform = "vst",
                                        interactor_column = "typeofcells",
                                        interest_column = "clinicaloutcome")

## converting counts to integer mode

## Error in checkFullRank(modelMatrix): the model matrix is not full rank, so the model cannot be fit as specified.
##   Levels or combinations of levels without any samples have resulted in
##   column(s) of zeros in the model matrix.
## 
##   Please read the vignette section 'Model matrix not full rank':
## 
##   vignette('DESeq2')

lrt_celltype_clinical_test[["favorite_genes"]]

## Error in eval(expr, envir, enclos): object 'lrt_celltype_clinical_test' not found

4.4.2 Look at only the differential genes

A good suggestion from Theresa was to examine only the most variant genes from failure vs. cure and see how they change the clustering/etc results. This is my attempt to address this query.

hs_clinic_topn <- sm(extract_significant_genes(hs_clinic_table, n = 100))
table <- "failure_vs_cure"
wanted <- rbind(hs_clinic_topn[["deseq"]][["ups"]][[table]],
                hs_clinic_topn[["deseq"]][["downs"]][[table]])

small_expt <- exclude_genes_expt(hs_clinical_nobiop, ids = rownames(wanted), method = "keep")

## Before removal, there were 19941 genes, now there are 200.

## There are 83 samples which kept less than 90 percent counts.

##  1017n1  1017m1  1034n1  1034n2  1034m2 1034m2-  2052e1  2052m2  2052n2  2052m3 
##  0.4344  0.4209  2.3933  2.8302  1.3082  1.2597  0.6909  0.6058  1.3510  0.6899 
##  2052n3  2065m1  2065n1  2066m1  2066n1  2065m2  2065n2  2065e2  2066m2  2066n2 
##  1.1861  1.1063  3.1233  0.4222  0.7710  0.6235  0.8864  1.3413  0.5375  0.9750 
##  2066e2  2068m1  2068n1  2068e1  2072m1  2072n1  2072e1  2071m1  2071n1  2073m1 
##  1.1821  0.5670  0.8161  1.3970  0.5170  0.6085  0.8754  0.7574  1.7531  0.6883 
##  2073n1  2073e1  2068m2  2068n2  2068e2  2072m2  2072n2  2072e2  2073m2  2073n2 
##  1.8493  0.9073  0.3942  0.5760  0.8514  0.4248  0.5737  0.6222  1.0724  2.7964 
##  2073e2  2066m3  2066n3  2066e3  2065e3  2068m3  2068n3  2068e3  2072m3  2072n3 
##  0.8723  0.4794  0.6782  0.6991  0.6026  0.4552  0.7345  0.8428  0.7395  1.9132 
##  2072e3  2073m3  2073n3  2073e3  2162m1  2162n1  2162e1  2162n2  2162e2  2162n3 
##  0.8600  0.5342  0.7511  0.6015  0.4937  0.6971  1.0733  0.9258  1.1809  0.7148 
##  2162e3  2167m1  2167n1  2167e1  2168m1  2168n1  2168e1  2168m2  2168n2  2168e2 
##  0.7263  0.6428  0.9420  1.3436  1.0901  2.2668  1.1321  0.9822  2.2914  0.9692 
##  2167m2  2167n3  2167e3  2168m3  2168n3  2168e3  2172n1  2172e1  1168m1  1168n1 
##  0.4135  1.4135  1.4795  1.3873  3.6918  1.1862  0.3999  0.5349  0.7780  1.3948 
##  1168m2  1168e2  1168n3 
##  0.7665  0.5925  0.9169

small_norm <- sm(normalize_expt(small_expt, transform = "log2", convert = "cpm",
                                norm = "quant", filter = TRUE))
plot_pca(small_norm)$plot

## Potentially check over the experimental design, there appear to be missing values.

## Warning in plot_pca(small_norm): There are NA values in the component data. This
## can lead to weird plotting errors.

## Warning: ggrepel: 17 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps

small_nb <- normalize_expt(small_expt, transform = "log2", convert = "cpm",
                           batch = "svaseq", norm = "quant", filter = TRUE)

## Warning in normalize_expt(small_expt, transform = "log2", convert = "cpm", :
## Quantile normalization and sva do not always play well together.

## Removing 0 low-count genes (200 remaining).

## batch_counts: Before batch/surrogate estimation, 1 entries are x==0: 0%.

## batch_counts: Before batch/surrogate estimation, 894 entries are 0<x<1: 5%.

## Setting 801 low elements to zero.

## transform_counts: Found 801 values equal to 0, adding 1 to the matrix.

plot_pca(small_nb)$plot

## Potentially check over the experimental design, there appear to be missing values.

## Warning in plot_pca(small_nb): There are NA values in the component data. This
## can lead to weird plotting errors.

## Warning: ggrepel: 79 unlabeled data points (too many overlaps). Consider
## increasing max.overlaps

## DESeq2 MA plot of failure / cure
hs_clinic_table[["plots"]][["failure_vs_cure"]][["deseq_ma_plots"]]$plot

## DESeq2 Volcano plot of failure / cure
hs_clinic_table[["plots"]][["failure_vs_cure"]][["deseq_vol_plots"]]$plot

4.4.3 g:Profiler results using the significant up and down genes

ups <- hs_clinic_sig[["deseq"]][["ups"]][[1]]
downs <- hs_clinic_sig[["deseq"]][["downs"]][[1]]

hs_clinic_gprofiler_ups <- simple_gprofiler(ups)

## Performing gProfiler GO search of 249 genes against hsapiens.

## GO search found 106 hits.

## Performing gProfiler KEGG search of 249 genes against hsapiens.

## KEGG search found 10 hits.

## Performing gProfiler REAC search of 249 genes against hsapiens.

## REAC search found 10 hits.

## Performing gProfiler MI search of 249 genes against hsapiens.

## MI search found 0 hits.

## Performing gProfiler TF search of 249 genes against hsapiens.

## TF search found 55 hits.

## Performing gProfiler CORUM search of 249 genes against hsapiens.

## CORUM search found 1 hits.

## Performing gProfiler HP search of 249 genes against hsapiens.

## HP search found 0 hits.

hs_clinic_gprofiler_ups[["pvalue_plots"]][["bpp_plot_over"]]

hs_clinic_gprofiler_ups[["pvalue_plots"]][["mfp_plot_over"]]

hs_clinic_gprofiler_ups[["pvalue_plots"]][["reactome_plot_over"]]

##hs_try2 <- simple_gprofiler2(ups)

hs_clinic_gprofiler_downs <- simple_gprofiler(downs)

## Performing gProfiler GO search of 232 genes against hsapiens.

## GO search found 58 hits.

## Performing gProfiler KEGG search of 232 genes against hsapiens.

## KEGG search found 8 hits.

## Performing gProfiler REAC search of 232 genes against hsapiens.

## REAC search found 5 hits.

## Performing gProfiler MI search of 232 genes against hsapiens.

## MI search found 0 hits.

## Performing gProfiler TF search of 232 genes against hsapiens.

## TF search found 11 hits.

## Performing gProfiler CORUM search of 232 genes against hsapiens.

## CORUM search found 0 hits.

## Performing gProfiler HP search of 232 genes against hsapiens.

## HP search found 2 hits.

hs_clinic_gprofiler_downs[["pvalue_plots"]][["bpp_plot_over"]]

hs_clinic_gprofiler_downs[["pvalue_plots"]][["mfp_plot_over"]]

hs_clinic_gprofiler_downs[["pvalue_plots"]][["reactome_plot_over"]]

4.5 Perform GSVA on the clinical samples

hs_celltype_gsva_c2 <- sm(simple_gsva(individual_celltypes))
hs_celltype_gsva_c2_sig <- sm(get_sig_gsva_categories(
    hs_celltype_gsva_c2,
    excel = "excel/individual_celltypes_gsva_c2.xlsx"))

broad_c7 <- GSEABase::getGmt("reference/msigdb/c7.all.v7.2.entrez.gmt",
                             collectionType = GSEABase::BroadCollection(category = "c7"),
                             geneIdType = GSEABase::EntrezIdentifier())
hs_celltype_gsva_c7 <- sm(simple_gsva(individual_celltypes, signatures = broad_c7,
                                      msig_xml = "reference/msigdb_v7.2.xml", cores = 10))
hs_celltype_gsva_c7_sig <- sm(get_sig_gsva_categories(
    hs_celltype_gsva_c7,
    excel = "excel/individual_celltypes_gsva_c7.xlsx"))

4.5.1 Print some plots of the GSVA outputs

## The raw heatmap of the C2 values
hs_celltype_gsva_c2_sig[["raw_plot"]]

## The 'significance' scores of the C2 values
hs_celltype_gsva_c2_sig[["score_plot"]]

## The subset of scores for categories deemed significantly different.
hs_celltype_gsva_c2_sig[["subset_plot"]]

## The raw heatmap of the C7 values
hs_celltype_gsva_c7_sig[["raw_plot"]]

## The 'significance' scores of the C7 values
hs_celltype_gsva_c7_sig[["score_plot"]]

## The subset of scores for categories deemed significantly different.
hs_celltype_gsva_c7_sig[["subset_plot"]]

5 Individual Cell types

The following blocks split the samples into a few groups by sample type and look at the distributions between them.

5.1 Implementation details

Get top/bottom n genes for each cell type, using clinical outcome as the factor of interest. For the moment, use sva for the DE analysis. Provide cpms for the top/bottom n genes.

Start with top/bottom 200. Perform default logFC and p-value as well.

5.1.1 Shared contrasts

Here is the contrast we will use throughput, I am leaving open the option to add more.

keepers <- list(
  "fail_vs_cure" = c("failure", "cure"))

5.2 Monocytes

5.2.1 Evaluate Monocyte samples

mono <- subset_expt(hs_valid, subset = "typeofcells=='monocytes'") %>%
  set_expt_conditions(fact = "clinicaloutcome") %>%
  set_expt_batches(fact = "donor") %>%
  set_expt_colors(colors = chosen_colors)

## subset_expt(): There were 152, now there are 28 samples.

## FIXME set_expt_colors should speak up if there are mismatches here!!!

save_result <- save(mono, file = "rda/monocyte_expt.rda")
mono_norm <- normalize_expt(mono, convert = "cpm", filter = TRUE,
                            transform = "log2", norm = "quant")

## Removing 8906 low-count genes (11035 remaining).

## transform_counts: Found 9 values equal to 0, adding 1 to the matrix.

plt <- plot_pca(mono_norm, plot_labels = FALSE)$plot
pp(file = glue("images/mono_pca_normalized-v{ver}.pdf"), image = plt)

mono_nb <- normalize_expt(mono, convert = "cpm", filter = TRUE,
                          transform = "log2", batch = "svaseq")

## Removing 8906 low-count genes (11035 remaining).

## batch_counts: Before batch/surrogate estimation, 1433 entries are x==0: 0%.

## batch_counts: Before batch/surrogate estimation, 18423 entries are 0<x<1: 6%.

## Setting 512 low elements to zero.

## transform_counts: Found 512 values equal to 0, adding 1 to the matrix.

plt <- plot_pca(mono_nb, plot_labels = FALSE)$plot
pp(file = glue("images/mono_pca_normalized_batch-v{ver}.pdf"), image = plt)

5.2.2 DE of Monocyte samples

5.2.2.1 Without sva

mono_de <- sm(all_pairwise(mono, model_batch = FALSE, filter = TRUE))

mono_tables <- sm(combine_de_tables(
    mono_de, keepers = keepers,
    excel = glue::glue("excel/monocyte_clinical_all_tables-v{ver}.xlsx")))
written <- write_xlsx(data = mono_tables[["data"]][[1]],
                      excel = glue::glue("excel/monocyte_clinical_table-v{ver}.xlsx"))
mono_sig <- sm(extract_significant_genes(mono_tables, according_to = "deseq"))
written <- write_xlsx(data = mono_sig[["deseq"]][["ups"]][[1]],
                      excel = glue::glue("excel/monocyte_clinical_sigup-v{ver}.xlsx"))
written <- write_xlsx(data = mono_sig[["deseq"]][["downs"]][[1]],
                      excel = glue::glue("excel/monocyte_clinical_sigdown-v{ver}.xlsx"))

mono_pct_sig <- sm(extract_significant_genes(mono_tables, n = 200,
                                             lfc = NULL, p = NULL, according_to = "deseq"))
written <- write_xlsx(data = mono_pct_sig[["deseq"]][["ups"]][[1]],
                      excel = glue::glue("excel/monocyte_clinical_sigup_pct-v{ver}.xlsx"))
written <- write_xlsx(data = mono_pct_sig[["deseq"]][["downs"]][[1]],
                      excel = glue::glue("excel/monocyte_clinical_sigdown_pct-v{ver}.xlsx"))
mono_sig$summary_df

## data frame with 0 columns and 1 row

## Print out a table of the cpm values for other explorations.
mono_cpm <- sm(normalize_expt(mono, convert = "cpm"))
written <- write_xlsx(data = exprs(mono_cpm),
                      excel = glue::glue("excel/monocyte_cpm_before_batch-v{ver}.xlsx"))
mono_bcpm <- sm(normalize_expt(mono, filter = TRUE, convert = "cpm", batch = "svaseq"))
written <- write_xlsx(data = exprs(mono_bcpm),
                      excel = glue::glue("excel/monocyte_cpm_after_batch-v{ver}.xlsx"))

5.2.2.2 With sva

mono_de_sva <- sm(all_pairwise(mono, model_batch = "svaseq", filter = TRUE))

mono_tables_sva <- sm(combine_de_tables(
    mono_de_sva, keepers = keepers,
    excel = glue::glue("excel/monocyte_clinical_all_tables_sva-v{ver}.xlsx")))
mono_sig_sva <- sm(extract_significant_genes(
    mono_tables_sva,
    excel = glue::glue("excel/monocyte_clinical_sig_tables_sva-v{ver}.xlsx"),
    according_to = "deseq"))

5.2.2.3 Monocyte DE plots

First print out the DE plots without and then with sva estimates.

## DESeq2 MA plot of failure / cure
mono_tables[["plots"]][["fail_vs_cure"]][["deseq_ma_plots"]]$plot

## DESeq2 Volcano plot of failure / cure
mono_tables[["plots"]][["fail_vs_cure"]][["deseq_vol_plots"]]$plot

## DESeq2 MA plot of failure / cure with svaseq
mono_tables_sva[["plots"]][["fail_vs_cure"]][["deseq_ma_plots"]]$plot

## DESeq2 Volcano plot of failure / cure with svaseq
mono_tables_sva[["plots"]][["fail_vs_cure"]][["deseq_vol_plots"]]$plot

5.2.2.4 Monocyte ontology search

ups <- mono_sig[["deseq"]][["ups"]][["fail_vs_cure"]]
downs <- mono_sig[["deseq"]][["downs"]][["fail_vs_cure"]]

mono_up_gp <- simple_gprofiler(sig_genes = ups)

## Performing gProfiler GO search of 126 genes against hsapiens.

## GO search found 32 hits.

## Performing gProfiler KEGG search of 126 genes against hsapiens.

## KEGG search found 6 hits.

## Performing gProfiler REAC search of 126 genes against hsapiens.

## REAC search found 3 hits.

## Performing gProfiler MI search of 126 genes against hsapiens.

## MI search found 0 hits.

## Performing gProfiler TF search of 126 genes against hsapiens.

## TF search found 17 hits.

## Performing gProfiler CORUM search of 126 genes against hsapiens.

## CORUM search found 1 hits.

## Performing gProfiler HP search of 126 genes against hsapiens.

## HP search found 20 hits.

mono_up_gp[["pvalue_plots"]][["bpp_plot_over"]]

mono_up_gp[["pvalue_plots"]][["mfp_plot_over"]]

mono_up_gp[["pvalue_plots"]][["reactome_plot_over"]]

mono_down_gp <- simple_gprofiler(sig_genes = downs)

## Performing gProfiler GO search of 281 genes against hsapiens.

## GO search found 98 hits.

## Performing gProfiler KEGG search of 281 genes against hsapiens.

## KEGG search found 1 hits.

## Performing gProfiler REAC search of 281 genes against hsapiens.

## REAC search found 8 hits.

## Performing gProfiler MI search of 281 genes against hsapiens.

## MI search found 0 hits.

## Performing gProfiler TF search of 281 genes against hsapiens.

## TF search found 126 hits.

## Performing gProfiler CORUM search of 281 genes against hsapiens.

## CORUM search found 2 hits.

## Performing gProfiler HP search of 281 genes against hsapiens.

## HP search found 1 hits.

mono_down_gp[["pvalue_plots"]][["bpp_plot_over"]]

mono_down_gp[["pvalue_plots"]][["mfp_plot_over"]]

mono_down_gp[["pvalue_plots"]][["reactome_plot_over"]]

5.2.2.5 Monocyte MSigDB query

broad_c7 <- GSEABase::getGmt("reference/msigdb/c7.all.v7.2.entrez.gmt",
                             collectionType = GSEABase::BroadCollection(category = "c7"),
                             geneIdType = GSEABase::EntrezIdentifier())

mono_up_goseq_msig <- goseq_msigdb(sig_genes = ups, signatures = broad_c7,
                                   signature_category = "c7", length_db = hs_length)

## Before conversion: 126, after conversion: 127.

## Before conversion: 227921, after conversion: 35341.

## Found 122 go_db genes and 127 length_db genes out of 127.

mono_down_goseq_msig <- goseq_msigdb(sig_genes = downs, signatures = broad_c7,
                                     signature_category = "c7", length_db = hs_length)

## Before conversion: 281, after conversion: 280.

## Before conversion: 227921, after conversion: 35341.

## Found 269 go_db genes and 280 length_db genes out of 280.

5.2.2.6 Plot of similar experiments

## Monocyte genes with increased expression in the failed samples
## share genes with the following experiments
mono_up_goseq_msig[["pvalue_plots"]][["mfp_plot_over"]]

## Monocyte genes with increased expression in the cured samples
## share genes with the following experiments
mono_down_goseq_msig[["pvalue_plots"]][["mfp_plot_over"]]

5.2.3 Evaluate Neutrophil samples

neut <- subset_expt(hs_valid, subset = "typeofcells=='neutrophils'") %>%
  set_expt_conditions(fact = "clinicaloutcome") %>%
  set_expt_batches(fact = "donor") %>%
  set_expt_colors(colors = chosen_colors)

## subset_expt(): There were 152, now there are 31 samples.

save_result <- save(neut, file = "rda/neutrophil_expt.rda")
neut_norm <- sm(normalize_expt(neut, convert = "cpm", filter = TRUE, transform = "log2"))
plt <- plot_pca(neut_norm, plot_labels = FALSE)$plot
pp(file = glue("images/neut_pca_normalized-v{ver}.pdf"), image = plt)

neut_nb <- sm(normalize_expt(neut, convert = "cpm", filter = TRUE,
                             transform = "log2", batch = "svaseq"))
plt <- plot_pca(neut_nb, plot_labels = FALSE)$plot
pp(file = glue("images/neut_pca_normalized_svaseq-v{ver}.pdf"), image = plt)

5.2.4 DE of Netrophil samples

5.2.4.1 Without sva

neut_de <- sm(all_pairwise(neut, model_batch = FALSE, filter = TRUE))

neut_tables <- sm(combine_de_tables(
    neut_de, keepers = keepers,
    excel = glue::glue("excel/neutrophil_clinical_all_tables-v{ver}.xlsx")))
written <- write_xlsx(data = neut_tables[["data"]][[1]],
                      excel = glue::glue("excel/neutrophil_clinical_table-v{ver}.xlsx"))
neut_sig <- sm(extract_significant_genes(neut_tables, according_to = "deseq"))
written <- write_xlsx(data = neut_sig[["deseq"]][["ups"]][[1]],
                      excel = glue::glue("excel/neutrophil_clinical_sigup-v{ver}.xlsx"))
written <- write_xlsx(data = neut_sig[["deseq"]][["downs"]][[1]],
                      excel = glue::glue("excel/neutrophil_clinical_sigdown-v{ver}.xlsx"))

neut_pct_sig <- sm(extract_significant_genes(neut_tables, n = 200, lfc = NULL,
                                             p = NULL, according_to = "deseq"))
written <- write_xlsx(data = neut_sig[["deseq"]][["ups"]][[1]],
                      excel = glue::glue("excel/neutrophil_clinical_sigup_pct-v{ver}.xlsx"))
written <- write_xlsx(data = neut_sig[["deseq"]][["downs"]][[1]],
                      excel = glue::glue("excel/neutrophil_clinical_sigdown_pct-v{ver}.xlsx"))
neut_cpm <- sm(normalize_expt(neut, convert = "cpm"))
written <- write_xlsx(data = exprs(neut_cpm),
                      excel = glue::glue("excel/neutrophil_cpm_before_batch-v{ver}.xlsx"))
neut_bcpm <- sm(normalize_expt(neut, filter = TRUE, batch = "svaseq", convert = "cpm"))
written <- write_xlsx(data = exprs(neut_bcpm),
                      excel = glue::glue("excel/neutrophil_cpm_after_batch-v{ver}.xlsx"))

5.2.4.2 With sva

neut_de_sva <- sm(all_pairwise(neut, model_batch = "svaseq", filter = TRUE))

neut_tables_sva <- sm(combine_de_tables(
    neut_de_sva, keepers = keepers,
    excel = glue::glue("excel/neutrophil_clinical_all_tables_sva-v{ver}.xlsx")))
neut_sig_sva <- sm(extract_significant_genes(
    neut_tables_sva,
    excel = glue::glue("excel/neutrophil_clinical_sig_tables_sva-v{ver}.xlsx"),
    according_to = "deseq"))

5.2.4.3 Neutrophil DE plots

## DESeq2 MA plot of failure / cure
neut_tables[["plots"]][["fail_vs_cure"]][["deseq_ma_plots"]]$plot

## DESeq2 Volcano plot of failure / cure
neut_tables[["plots"]][["fail_vs_cure"]][["deseq_vol_plots"]]$plot

## DESeq2 MA plot of failure / cure with sva
neut_tables_sva[["plots"]][["fail_vs_cure"]][["deseq_ma_plots"]]$plot

## DESeq2 Volcano plot of failure / cure with sva
neut_tables_sva[["plots"]][["fail_vs_cure"]][["deseq_vol_plots"]]$plot

5.2.4.4 Neutrophil ontology search

ups <- neut_sig[["deseq"]][["ups"]][["fail_vs_cure"]]
downs <- neut_sig[["deseq"]][["downs"]][["fail_vs_cure"]]

neut_up_gp <- simple_gprofiler(sig_genes = ups)

## Performing gProfiler GO search of 179 genes against hsapiens.

## GO search found 41 hits.

## Performing gProfiler KEGG search of 179 genes against hsapiens.

## KEGG search found 3 hits.

## Performing gProfiler REAC search of 179 genes against hsapiens.

## REAC search found 5 hits.

## Performing gProfiler MI search of 179 genes against hsapiens.

## MI search found 0 hits.

## Performing gProfiler TF search of 179 genes against hsapiens.

## TF search found 35 hits.

## Performing gProfiler CORUM search of 179 genes against hsapiens.

## CORUM search found 0 hits.

## Performing gProfiler HP search of 179 genes against hsapiens.

## HP search found 4 hits.

neut_up_gp[["pvalue_plots"]][["bpp_plot_over"]]

neut_up_gp[["pvalue_plots"]][["mfp_plot_over"]]

neut_up_gp[["pvalue_plots"]][["reactome_plot_over"]]

neut_down_gp <- simple_gprofiler(downs)

## Performing gProfiler GO search of 138 genes against hsapiens.

## GO search found 0 hits.

## Performing gProfiler KEGG search of 138 genes against hsapiens.

## KEGG search found 0 hits.

## Performing gProfiler REAC search of 138 genes against hsapiens.

## REAC search found 0 hits.

## Performing gProfiler MI search of 138 genes against hsapiens.

## MI search found 0 hits.

## Performing gProfiler TF search of 138 genes against hsapiens.

## TF search found 2 hits.

## Performing gProfiler CORUM search of 138 genes against hsapiens.

## CORUM search found 0 hits.

## Performing gProfiler HP search of 138 genes against hsapiens.

## HP search found 0 hits.

neut_down_gp[["pvalue_plots"]][["bpp_plot_over"]]

## NULL

neut_down_gp[["pvalue_plots"]][["mfp_plot_over"]]

## NULL

neut_down_gp[["pvalue_plots"]][["reactome_plot_over"]]

## NULL

5.2.4.5 Neutrophil GSVA query

neut_up_goseq_msig <- goseq_msigdb(sig_genes = ups, signatures = broad_c7,
                                   signature_category = "c7", length_db = hs_length)

## Before conversion: 179, after conversion: 178.

## Before conversion: 227921, after conversion: 35341.

## Found 171 go_db genes and 178 length_db genes out of 178.

neut_down_goseq_msig <- goseq_msigdb(sig_genes = downs, signatures = broad_c7,
                                     signature_category = "c7", length_db = hs_length)

## Before conversion: 138, after conversion: 133.

## Before conversion: 227921, after conversion: 35341.

## Found 128 go_db genes and 133 length_db genes out of 133.

5.2.4.6 Plot of similar experiments

## Neutrophil genes with increased expression in the failed samples
## share genes with the following experiments
neut_up_goseq_msig[["pvalue_plots"]][["mfp_plot_over"]]

## Neutrophil genes with increased expression in the cured samples
## share genes with the following experiments
neut_down_goseq_msig[["pvalue_plots"]][["mfp_plot_over"]]

5.3 Eosinophils

5.3.1 Evaluate Eosinophil samples

eo <- subset_expt(hs_valid, subset = "typeofcells=='eosinophils'") %>%
  set_expt_conditions(fact = "clinicaloutcome") %>%
  set_expt_batches(fact = "donor") %>%
  set_expt_colors(colors = chosen_colors)

## subset_expt(): There were 152, now there are 24 samples.

save_result <- save(eo, file = "rda/eosinophil_expt.rda")
eo_norm <- sm(normalize_expt(eo, convert = "cpm", transform = "log2",
                             norm = "quant", filter = TRUE))
plt <- plot_pca(eo_norm, plot_labels = FALSE)$plot
pp(file = glue("images/eo_pca_normalized-v{ver}.pdf"), image = plt)

eo_nb <- sm(normalize_expt(eo, convert = "cpm", transform = "log2",
                           filter = TRUE, batch = "svaseq"))
plot_pca(eo_nb)$plot

5.3.2 DE of Eosinophil samples

5.3.2.1 Withouth sva

eo_de <- sm(all_pairwise(eo, model_batch = FALSE, filter = TRUE))

eo_tables <- sm(combine_de_tables(
    eo_de, keepers = keepers,
    excel = glue::glue("excel/eosinophil_clinical_all_tables-v{ver}.xlsx")))
written <- write_xlsx(data = eo_tables[["data"]][[1]],
                      excel = glue::glue("excel/eosinophil_clinical_table-v{ver}.xlsx"))
eo_sig <- sm(extract_significant_genes(eo_tables, according_to = "deseq"))
written <- write_xlsx(data = eo_sig[["deseq"]][["ups"]][[1]],
                      excel = glue::glue("excel/eosinophil_clinical_sigup-v{ver}.xlsx"))
written <- write_xlsx(data = eo_sig[["deseq"]][["downs"]][[1]],
                      excel = glue::glue("excel/eosinophil_clinical_sigdown-v{ver}.xlsx"))

eo_pct_sig <- sm(extract_significant_genes(eo_tables, n = 200,
                                           lfc = NULL, p = NULL, according_to = "deseq"))
written <- write_xlsx(data = eo_pct_sig[["deseq"]][["ups"]][[1]],
                      excel = glue::glue("excel/eosinophil_clinical_sigup_pct-v{ver}.xlsx"))
written <- write_xlsx(data = eo_pct_sig[["deseq"]][["downs"]][[1]],
                      excel = glue::glue("excel/eosinophil_clinical_sigdown_pct-v{ver}.xlsx"))

eo_cpm <- sm(normalize_expt(eo, convert = "cpm"))
written <- write_xlsx(data = exprs(eo_cpm),
                      excel = glue::glue("excel/eosinophil_cpm_before_batch-v{ver}.xlsx"))
eo_bcpm <- sm(normalize_expt(eo, filter = TRUE, batch = "svaseq", convert = "cpm"))
written <- write_xlsx(data = exprs(eo_bcpm),
                      excel = glue::glue("excel/eosinophil_cpm_after_batch-v{ver}.xlsx"))

5.3.2.2 With sva

eo_de_sva <- sm(all_pairwise(eo, model_batch = "svaseq", filter = TRUE))

eo_tables_sva <- sm(combine_de_tables(
    eo_de_sva, keepers = keepers,
    excel = glue::glue("excel/eosinophil_clinical_all_tables_sva-v{ver}.xlsx")))
eo_sig_sva <- sm(extract_significant_genes(
    eo_tables_sva,
    excel = glue::glue("excel/eosinophil_clinical_sig_tables_sva-v{ver}.xlsx"),
    according_to = "deseq"))

5.3.2.3 Eosinophil DE plots

## DESeq2 MA plot of failure / cure
eo_tables[["plots"]][["fail_vs_cure"]][["deseq_ma_plots"]]$plot

## DESeq2 Volcano plot of failure / cure
eo_tables[["plots"]][["fail_vs_cure"]][["deseq_vol_plots"]]$plot

## DESeq2 MA plot of failure / cure with sva
eo_tables_sva[["plots"]][["fail_vs_cure"]][["deseq_ma_plots"]]$plot

## DESeq2 Volcano plot of failure / cure with sva
eo_tables_sva[["plots"]][["fail_vs_cure"]][["deseq_vol_plots"]]$plot

5.3.2.4 Eosinophil ontology search

ups <- eo_sig[["deseq"]][["ups"]][["fail_vs_cure"]]
downs <- eo_sig[["deseq"]][["downs"]][["fail_vs_cure"]]

eo_up_gp <- simple_gprofiler(sig_genes = ups)

## Performing gProfiler GO search of 110 genes against hsapiens.

## GO search found 102 hits.

## Performing gProfiler KEGG search of 110 genes against hsapiens.

## KEGG search found 7 hits.

## Performing gProfiler REAC search of 110 genes against hsapiens.

## REAC search found 9 hits.

## Performing gProfiler MI search of 110 genes against hsapiens.

## MI search found 0 hits.

## Performing gProfiler TF search of 110 genes against hsapiens.

## TF search found 50 hits.

## Performing gProfiler CORUM search of 110 genes against hsapiens.

## CORUM search found 0 hits.

## Performing gProfiler HP search of 110 genes against hsapiens.

## HP search found 0 hits.

eo_up_gp[["pvalue_plots"]][["bpp_plot_over"]]

eo_up_gp[["pvalue_plots"]][["mfp_plot_over"]]

eo_up_gp[["pvalue_plots"]][["reactome_plot_over"]]

eo_down_gp <- simple_gprofiler(downs)

## Performing gProfiler GO search of 38 genes against hsapiens.

## GO search found 0 hits.

## Performing gProfiler KEGG search of 38 genes against hsapiens.

## KEGG search found 0 hits.

## Performing gProfiler REAC search of 38 genes against hsapiens.

## REAC search found 0 hits.

## Performing gProfiler MI search of 38 genes against hsapiens.

## MI search found 0 hits.

## Performing gProfiler TF search of 38 genes against hsapiens.

## TF search found 0 hits.

## Performing gProfiler CORUM search of 38 genes against hsapiens.

## CORUM search found 0 hits.

## Performing gProfiler HP search of 38 genes against hsapiens.

## HP search found 0 hits.

eo_down_gp[["pvalue_plots"]][["bpp_plot_over"]]

## NULL

eo_down_gp[["pvalue_plots"]][["mfp_plot_over"]]

## NULL

eo_down_gp[["pvalue_plots"]][["reactome_plot_over"]]

## NULL

5.3.2.5 Eosinophil MSigDB query

eo_up_goseq_msig <- goseq_msigdb(sig_genes = ups, signatures = broad_c7,
                                 signature_category = "c7", length_db = hs_length)

## Before conversion: 110, after conversion: 110.

## Before conversion: 227921, after conversion: 35341.

## Found 108 go_db genes and 110 length_db genes out of 110.

eo_down_goseq_msig <- goseq_msigdb(sig_genes = downs, signatures = broad_c7,
                                   signature_category = "c7", length_db = hs_length)

## Before conversion: 38, after conversion: 37.

## Before conversion: 227921, after conversion: 35341.

## Found 33 go_db genes and 37 length_db genes out of 37.

5.3.2.6 Plot of similar experiments

## Eosinophil genes with increased expression in the failed samples
## share genes with the following experiments
eo_up_goseq_msig[["pvalue_plots"]][["mfp_plot_over"]]

## Eosinophil genes with increased expression in the cured samples
## share genes with the following experiments
eo_down_goseq_msig[["pvalue_plots"]][["mfp_plot_over"]]

5.4 Biopsies

5.4.1 Evaluate Biopsy samples

biop <- subset_expt(hs_valid, subset = "typeofcells=='biopsy'") %>%
  set_expt_conditions(fact = "clinicaloutcome") %>%
  set_expt_batches(fact = "donor") %>%
  set_expt_colors(colors = chosen_colors)

## subset_expt(): There were 152, now there are 49 samples.

save_result <- save(biop, file = "rda/biopsy_expt.rda")
biop_norm <- normalize_expt(biop, filter = TRUE, convert = "cpm",
                            transform = "log2", norm = "quant")

## Removing 5753 low-count genes (14188 remaining).

## transform_counts: Found 16 values equal to 0, adding 1 to the matrix.

plt <- plot_pca(biop_norm, plot_labels = FALSE)$plot

## Error in data.frame(sampleid = as.character(design[["sampleid"]]), condition = design[[cond_column]], : arguments imply differing number of rows: 49, 48

pp(file = glue("images/biop_pca_normalized-v{ver}.pdf"), image = plt)

biop_nb <- sm(normalize_expt(biop, convert = "cpm", filter = TRUE,
                             transform = "log2", batch = "svaseq"))
plt <- plot_pca(biop_nb, plot_labels = FALSE)$plot

## Error in data.frame(sampleid = as.character(design[["sampleid"]]), condition = design[[cond_column]], : arguments imply differing number of rows: 49, 48

pp(file = glue("images/biop_pca_normalized_svaseq-v{ver}.pdf"), image = plt)

5.4.2 DE of Biopsy samples

5.4.2.1 Without sva

biop_de <- sm(all_pairwise(biop, model_batch = FALSE, filter = TRUE))

## Error in data.frame(sampleid = as.character(design[["sampleid"]]), condition = design[[cond_column]], : arguments imply differing number of rows: 49, 48

biop_tables <- combine_de_tables(biop_de, keepers = keepers,
                                 excel = glue::glue("excel/biopsy_clinical_all_tables-v{ver}.xlsx"))

## Deleting the file excel/biopsy_clinical_all_tables-v202106.xlsx before writing the tables.

## Error in combine_de_tables(biop_de, keepers = keepers, excel = glue::glue("excel/biopsy_clinical_all_tables-v{ver}.xlsx")): object 'biop_de' not found

written <- write_xlsx(data = biop_tables[["data"]][[1]],
                      excel = glue::glue("excel/biopsy_clinical_table-v{ver}.xlsx"))

## Error in write_xlsx(data = biop_tables[["data"]][[1]], excel = glue::glue("excel/biopsy_clinical_table-v{ver}.xlsx")): object 'biop_tables' not found

biop_sig <- extract_significant_genes(biop_tables, according_to = "deseq")

## Error in extract_significant_genes(biop_tables, according_to = "deseq"): object 'biop_tables' not found

##written <- write_xlsx(data = biop_sig[["deseq"]][["ups"]][[1]],
##                      excel = glue::glue("excel/biopsy_clinical_sigup-v{ver}.xlsx"))
written <- write_xlsx(data = biop_sig[["deseq"]][["downs"]][[1]],
                      excel = glue::glue("excel/biopsy_clinical_sigdown-v{ver}.xlsx"))

## Error in write_xlsx(data = biop_sig[["deseq"]][["downs"]][[1]], excel = glue::glue("excel/biopsy_clinical_sigdown-v{ver}.xlsx")): object 'biop_sig' not found

biop_pct_sig <- extract_significant_genes(biop_tables, n = 200, lfc = NULL, p = NULL, according_to = "deseq")

## Error in extract_significant_genes(biop_tables, n = 200, lfc = NULL, p = NULL, : object 'biop_tables' not found

written <- write_xlsx(data = biop_pct_sig[["deseq"]][["ups"]][[1]],
                      excel = glue::glue("excel/biopsy_clinical_sigup_pct-v{ver}.xlsx"))

## Error in write_xlsx(data = biop_pct_sig[["deseq"]][["ups"]][[1]], excel = glue::glue("excel/biopsy_clinical_sigup_pct-v{ver}.xlsx")): object 'biop_pct_sig' not found

written <- write_xlsx(data = biop_pct_sig[["deseq"]][["downs"]][[1]],
                      excel = glue::glue("excel/biopsy_clinical_sigdown_pct-v{ver}.xlsx"))

## Error in write_xlsx(data = biop_pct_sig[["deseq"]][["downs"]][[1]], excel = glue::glue("excel/biopsy_clinical_sigdown_pct-v{ver}.xlsx")): object 'biop_pct_sig' not found

biop_cpm <- sm(normalize_expt(biop, convert = "cpm"))
written <- write_xlsx(data = exprs(biop_cpm),
                      excel = glue::glue("excel/biopsy_cpm_before_batch-v{ver}.xlsx"))
biop_bcpm <- sm(normalize_expt(biop, filter = TRUE, batch = "svaseq", convert = "cpm"))
written <- write_xlsx(data = exprs(biop_bcpm),
                      excel = glue::glue("excel/biopsy_cpm_after_batch-v{ver}.xlsx"))

5.4.2.2 with sva

biop_de_sva <- sm(all_pairwise(biop, model_batch = "svaseq", filter = TRUE))

## Error in data.frame(sample = samples, factor = as.integer(as.factor(expt[["design"]][[batch_column]])), : arguments imply differing number of rows: 49, 48, 1

biop_tables_sva <- sm(combine_de_tables(
    biop_de_sva, keepers = keepers,
    excel = glue::glue("excel/biopsy_clinical_all_tables_sva-v{ver}.xlsx")))

## Error in combine_de_tables(biop_de_sva, keepers = keepers, excel = glue::glue("excel/biopsy_clinical_all_tables_sva-v{ver}.xlsx")): object 'biop_de_sva' not found

biop_sig_sva <- sm(extract_significant_genes(
    biop_tables_sva,
    excel = glue::glue("excel/biopsy_clinical_sig_tables_sva-v{ver}.xlsx"),
    according_to = "deseq"))

## Error in extract_significant_genes(biop_tables_sva, excel = glue::glue("excel/biopsy_clinical_sig_tables_sva-v{ver}.xlsx"), : object 'biop_tables_sva' not found

5.4.2.3 Biopsy DE plots

## DESeq2 MA plot of failure / cure
biop_tables[["plots"]][["fail_vs_cure"]][["deseq_ma_plots"]]$plot

## Error in eval(expr, envir, enclos): object 'biop_tables' not found

## DESeq2 Volcano plot of failure / cure
biop_tables[["plots"]][["fail_vs_cure"]][["deseq_vol_plots"]]$plot

## Error in eval(expr, envir, enclos): object 'biop_tables' not found

## DESeq2 MA plot of failure / cure
biop_tables_sva[["plots"]][["fail_vs_cure"]][["deseq_ma_plots"]]$plot

## Error in eval(expr, envir, enclos): object 'biop_tables_sva' not found

## DESeq2 Volcano plot of failure / cure
biop_tables_sva[["plots"]][["fail_vs_cure"]][["deseq_vol_plots"]]$plot

## Error in eval(expr, envir, enclos): object 'biop_tables_sva' not found

6 Look for shared genes among Monocytes/Neutrophils/Eosinophils

We have three variables containing the ‘significant’ DE genes for the three cell types. For this I am choosing (for the moment) to use the sva data.

## mono_sig_sva, neut_sig_sva, eo_sig_sva
sig_vectors <- list(
    "monocytes" = c(rownames(mono_sig_sva[["deseq"]][["ups"]][["fail_vs_cure"]]),
                    rownames(mono_sig_sva[["deseq"]][["downs"]][["fail_vs_cure"]])),
    "neutrophils" = c(rownames(neut_sig_sva[["deseq"]][["ups"]][["fail_vs_cure"]]),
                      rownames(neut_sig_sva[["deseq"]][["downs"]][["fail_vs_cure"]])),
    "eosinophils" =  c(rownames(eo_sig_sva[["deseq"]][["ups"]][["fail_vs_cure"]]),
                       rownames(eo_sig_sva[["deseq"]][["downs"]][["fail_vs_cure"]])))

shared_vector <- Vennerable::Venn(Sets = sig_vectors)
Vennerable::plot(shared_vector, doWeights = FALSE)

shared_ids <- shared_vector@IntersectionSets[["111"]]
shared_expt <- exclude_genes_expt(hs_clinical, ids = shared_ids, method = "keep")

## Before removal, there were 19941 genes, now there are 21.

## There are 132 samples which kept less than 90 percent counts.

##  1017n1  1017m1  1034n1 1034bp1  1034n2  1034m2 1034m2- 2050bp1 2052bp1  2052e1 
## 0.10770 0.10087 0.95547 0.33773 1.09654 0.51353 0.49517 0.16070 0.32817 0.03808 
##  2052m2  2052n2  2052m3  2052n3  2065m1  2065n1 2065bp1  2066m1  2066n1 2066bp1 
## 0.12555 0.24676 0.16692 0.23432 0.16654 0.52827 0.34346 0.07692 0.13150 0.27139 
##  2065m2  2065n2  2065e2  2066m2  2066n2  2066e2  2068m1  2068n1  2068e1 2068bp1 
## 0.07087 0.04962 0.03702 0.06761 0.06831 0.02495 0.04981 0.04888 0.01947 0.17676 
##  2072m1  2072n1  2072e1 2072bp1  2071m1  2071n1 2071bp1  2073m1  2073n1  2073e1 
## 0.06351 0.06770 0.03133 0.18845 0.12420 0.39016 0.23609 0.13451 0.45494 0.04816 
## 2073bp1  2068m2  2068n2  2068e2  2072m2  2072n2  2072e2  2073m2  2073n2  2073e2 
## 0.25557 0.06535 0.07704 0.01960 0.09097 0.11434 0.03699 0.27127 0.94140 0.12439 
##  2066m3  2066n3  2066e3  2065e3  2068m3  2068n3  2068e3  2072m3  2072n3  2072e3 
## 0.08602 0.10998 0.02804 0.02263 0.06194 0.06636 0.02747 0.18905 0.57829 0.06830 
## 2159bp1  2073m3  2073n3  2073e3  2162m1  2162n1  2162e1 2162bp1  2162n2  2162e2 
## 0.21495 0.13146 0.21920 0.05208 0.06606 0.06570 0.04768 0.21141 0.08682 0.07029 
##  2162n3  2162e3  2167m1  2167n1  2167e1 2167bp1  2168m1  2168n1  2168e1  2168m2 
## 0.13566 0.04089 0.13729 0.06095 0.11876 0.18832 0.32153 0.76205 0.16761 0.26397 
##  2168n2  2168e2  2167m2  2167n3  2167e3  2168m3  2168n3  2168e3  2172n1  2172e1 
## 0.75628 0.13855 0.07451 0.06879 0.08244 0.35171 0.98796 0.18522 0.06621 0.03115 
## 2172bp1 2173bp1 1154bp1  2008-1  2008-2  2008-3  1029-1  1029-2  1029-3  1036-1 
## 0.34453 0.23774 0.23862 0.18147 0.13122 0.19344 0.18687 0.17736 0.15556 0.29831 
##  1036-2  1036-3  1037-1  1037-2  1037-3  1031-1  1031-2  1031-3  2002-1  2002-3 
## 0.23483 0.18661 0.28122 0.13634 0.14125 0.30709 0.16886 0.19346 0.26122 0.12688 
##  2009-1  2009-2  2010-1  2010-2  2010-3  1019-1  1019-2  1019-3  2004-1  2004-2 
## 0.19448 0.14828 0.24210 0.24601 0.15844 0.41003 0.27201 0.23768 0.17984 0.17389 
##  2004-3  2001-1  2001-2  2001-3  2003-1  2003-2  2003-3  1168m1  1168n1  1168m2 
## 0.14816 0.24701 0.15282 0.10547 0.39666 0.16892 0.14402 0.24770 0.39685 0.23465 
##  1168e2  1168n3 
## 0.05943 0.27880

shared_written <- write_expt(shared_expt,
                             excel = glue::glue("excel/shared_across_celltypes-v{ver}.xlsx"))

## Writing the first sheet, containing a legend and some summary data.

## Error in data.frame(id = colnames(data), nonzero_genes = colSums(data >=  : 
##   arguments imply differing number of rows: 132, 131
## Error in data.frame(id = colnames(mtrx), sum = colSums(mtrx), condition = condition,  : 
##   arguments imply differing number of rows: 132, 131
## Error in heatmap.3(heatmap_data, keysize = keysize, labRow = expt_names,  : 
##   'ColSideColors' must be a character vector of length ncol(x)
## Error in data.frame(sample = rownames(properties), sm = prop_median, condition = conditions,  : 
##   arguments imply differing number of rows: 132, 131
## Error in heatmap.3(heatmap_data, keysize = keysize, labRow = expt_names,  : 
##   'ColSideColors' must be a character vector of length ncol(x)
## Error in data.frame(sample = rownames(properties), sm = prop_median, condition = conditions,  : 
##   arguments imply differing number of rows: 132, 131
## Error in data.frame(sampleid = as.character(design[["sampleid"]]), condition = design[[cond_column]],  : 
##   arguments imply differing number of rows: 132, 131
## Error in data.frame(sampleid = as.character(design[["sampleid"]]), condition = design[[cond_column]],  : 
##   arguments imply differing number of rows: 132, 131
## Error in names(colors) <- design[, "condition"] : 
##   'names' attribute [132] must be the same length as the vector [131]
## Error in data.frame(sampleid = as.character(design[["sampleid"]]), condition = design[[cond_column]],  : 
##   arguments imply differing number of rows: 132, 131

## Error in `[.data.frame`(as.data.frame(metrics[["libsizes"]]), , c("id", : undefined columns selected

7 Monocytes by visit

1. Can you please share with us a PCA (SVA and non-SVA) of the monocytes of the TMRC3 project, but labeling them based on the visit (V1, V2, V3)?
2. Can you please share DE lists of V1 vs V2, V1 vs V3, V1 vs. V2+V3 and V2 vs V3?

visit_colors <- chosen_colors <- c("#D95F02", "#7570B3", "#1B9E77")
names(visit_colors) <- c(1, 2, 3)
mono_visit <- subset_expt(hs_valid, subset = "typeofcells=='monocytes'") %>%
  set_expt_conditions(fact = "visitnumber") %>%
  set_expt_batches(fact = "clinicaloutcome") %>%
  set_expt_colors(colors = chosen_colors)

## subset_expt(): There were 152, now there are 28 samples.

mono_visit_norm <- normalize_expt(mono_visit, filter = TRUE, norm = "quant", convert = "cpm",
                                  transform = "log2")

## Removing 8906 low-count genes (11035 remaining).

## transform_counts: Found 9 values equal to 0, adding 1 to the matrix.

mono_visit_pca <- plot_pca(mono_visit_norm)
pp(file = "images/monocyte_by_visit.png", image = mono_visit_pca$plot)

mono_visit_nb <- normalize_expt(mono_visit, filter = TRUE, convert = "cpm",
                                batch = "svaseq", transform = "log2")

## Removing 8906 low-count genes (11035 remaining).

## batch_counts: Before batch/surrogate estimation, 1433 entries are x==0: 0%.

## batch_counts: Before batch/surrogate estimation, 18423 entries are 0<x<1: 6%.

## Setting 401 low elements to zero.

## transform_counts: Found 401 values equal to 0, adding 1 to the matrix.

mono_visit_nb_pca <- plot_pca(mono_visit_nb)
pp(file = "images/monocyte_by_visit_nb.png", image = mono_visit_nb_pca$plot)

table(pData(mono_visit_norm)$batch)

## 
##    cure failure    lost 
##       9      13       6

keepers <- list(
    "second_vs_first" = c("c2", "c1"),
    "third_vs_second" = c("c3", "c2"),
    "third_vs_first" = c("c3", "c1"))
mono_visit_de <- all_pairwise(mono_visit, model_batch = "svaseq", filter = TRUE)

## batch_counts: Before batch/surrogate estimation, 1433 entries are x==0: 0%.

## Plotting a PCA before surrogate/batch inclusion.

## Using svaseq to visualize before/after batch inclusion.

## Performing a test normalization with: raw

## Removing 0 low-count genes (11035 remaining).

## batch_counts: Before batch/surrogate estimation, 1433 entries are x==0: 0%.

## batch_counts: Before batch/surrogate estimation, 18423 entries are 0<x<1: 6%.

## Setting 401 low elements to zero.

## transform_counts: Found 401 values equal to 0, adding 1 to the matrix.

## Finished running DE analyses, collecting outputs.

## Comparing analyses.

mono_visit_tables <- combine_de_tables(
    mono_visit_de,
    keepers = keepers,
    excel = glue::glue("excel/mono_visit_tables-v{ver}.xlsx"))

## Deleting the file excel/mono_visit_tables-v202106.xlsx before writing the tables.

new_factor <- as.character(pData(mono_visit)[["visitnumber"]])
not_one_idx <- new_factor != 1
new_factor[not_one_idx] <- "not_1"
mono_one_vs <- set_expt_conditions(mono_visit, new_factor)

mono_one_vs_de <- all_pairwise(mono_one_vs, model_batch = "svaseq", filter = TRUE)

## batch_counts: Before batch/surrogate estimation, 1433 entries are x==0: 0%.

## Plotting a PCA before surrogate/batch inclusion.

## Using svaseq to visualize before/after batch inclusion.

## Performing a test normalization with: raw

## Removing 0 low-count genes (11035 remaining).

## batch_counts: Before batch/surrogate estimation, 1433 entries are x==0: 0%.

## batch_counts: Before batch/surrogate estimation, 18423 entries are 0<x<1: 6%.

## Setting 384 low elements to zero.

## transform_counts: Found 384 values equal to 0, adding 1 to the matrix.

## Finished running DE analyses, collecting outputs.

## Comparing analyses.

mono_one_vs_tables <- combine_de_tables(
    mono_one_vs_de,
    excel = glue::glue("excel/mono_one_vs_tables-v{ver}.xlsx"))

## Deleting the file excel/mono_one_vs_tables-v202106.xlsx before writing the tables.

8 Test TSP

In writing the following, I quickly realized that tspair was not joking when it said it is intended for small numbers of genes. For a full expressionset of human data it is struggling. I like the idea, it may prove worth while to spend some time optimizing the package so that it is more usable.

expt <- hs_clinical_nobiop

simple_tsp <- function(expt, column = "condition") {
  facts <- levels(as.factor(pData(expt)[[column]]))
  retlist <- list()
  if (length(facts) < 2) {
    stop("This requires factors with at least 2 levels.")
  } else if (length(facts) == 2) {
    retlist <- simple_tsp_pair(expt, column = column)
  } else {
    for (first in 1:(length(facts) - 1)) {
      for (second in 2:(length(facts))) {
        if (first < second) {
          name <- glue::glue("{facts[first]}_vs_{facts[second]}")
          message("Starting ", name, ".")
          substring <- glue::glue("{column}=='{facts[first]}'|{column}=='{facts[second]}'")
          subby <- subset_expt(expt, subset=as.character(substring))
          retlist[[name]] <- simple_tsp_pair(subby, column = column)
        }
      }
    }
  }
}

simple_tsp_pair <- function(subby, column = "condition", repetitions = 50) {
  tsp_input <- subby[["expressionset"]]
  tsp_output <- tspcalc(tsp_input, column)
  tsp_scores <- tspsig(tsp_input, column, B = repetitions)
}

tsp1 <- tspcalc(tsp_input, "condition")

if (!isTRUE(get0("skip_load"))) {
  pander::pander(sessionInfo())
  message(paste0("This is hpgltools commit: ", get_git_commit()))
  message(paste0("Saving to ", savefile))
  tmp <- sm(saveme(filename = savefile))
}

## If you wish to reproduce this exact build of hpgltools, invoke the following:

## > git clone http://github.com/abelew/hpgltools.git

## > git reset 72947fcc6afe09da22d71967059edd84e3063341

## This is hpgltools commit: Tue Jun 1 15:57:56 2021 -0400: 72947fcc6afe09da22d71967059edd84e3063341

## Saving to tmrc3_02sample_estimation_v202106.rda.xz

tmp <- loadme(filename = savefile)