DESeq2 iterate as an alternative sizeFactor estimator
1
0
Entering edit mode
Lauma R ▴ 10
@lauma-r-14513
Last seen 6.1 years ago
Cambridge

Dear All, 

 

I am doing an analysis on human  primary cell samples +/- treatment.  Due shortage of cells we constructed the library from pg RNA  as starting material plus the sequencing returned very low counts for most of genes. 

I  extracted the protein coding genes from the matrix . However,  the default method for size factor estimation returns NULL.  At the begging I thought it is because my median counts is 0 so geom mean would not work, but after I filtered out genes (more than 7000) that had counts across all samples it still gives NULL. 

 I tried other methods suggested and iterate works well; I know the " "iterate" estimator iterates between estimating the dispersion with a design of ~1, and finding a size factor vector by numerically optimizing the likelihood of the ~1 model" , but I do not think that I clearly understand what exactly it does to my data. 

In addition I get very high dispersion estimates.

 

deseq2 estimatesizefactors • 1.7k views
ADD COMMENT
0
Entering edit mode

> dds<- DESeqDataSetFromTximport(txi_simplesum, sampleTable, ~ 1)
using counts and average transcript lengths from tximport

> dds.protein.coding <- dds[rownames(dds) %in% Sample_coding.v]
> dds.protein.coding
class: DESeqDataSet 
dim: 18762 94 
metadata(1): version
assays(2): counts avgTxLength
rownames(18762): ENSG00000000419 ENSG00000000457 ... ENSG00000280267 ENSG00000280433
rowData names(0):
colnames(94): 0406_S18_S48_L008 0406_S19_S26_L005 ... 2105S2_S30 3007S6_S29
colData names(18): Sampl.name RUN ... SAMPLE.TYPE Condition
> Blood.CD19.coding <- dds.protein.coding[ ,dds.protein.coding$Cell.type %in% "BLOOD.CD19"]
> Blood.CD19.coding
class: DESeqDataSet 
dim: 18762 26 
metadata(1): version
assays(2): counts avgTxLength
rownames(18762): ENSG00000000419 ENSG00000000457 ... ENSG00000280267 ENSG00000280433
rowData names(0):
colnames(26): 07_02_2017_6D_S1_L001 1405_S6_S36_L006 ... 3004_S6_S17_L004 3007S6_S29
colData names(18): Sampl.name RUN ... SAMPLE.TYPE Condition
> design(Blood.CD19.coding) <- ~ Condition
> GeneCounts.Blood.CD19 <- counts(Blood.CD19.coding)
> idx.nz.Blood.CD19 <- apply(GeneCounts.Blood.CD19, 1, function(x) { all(x >0)})
> sum(idx.nz.Blood.CD19)
[1] 7749
> Blood.CD19.ratio <- estimateSizeFactors(Blood.CD19.coding) 
using 'avgTxLength' from assays(dds), correcting for library size
> sizeFactors(Blood.CD19.ratio)
NULL
> Blood.CD19.iterate <- estimateSizeFactors(Blood.CD19.coding, type = ("iterate"))
> sizeFactors(Blood.CD19.iterate)
 07_02_2017_6D_S1_L001       1405_S6_S36_L006       1412_S6_S17_L003       1606_S6_S21_L004 
             1.7455888              0.5103565              1.2959369              0.7646775 
      2001_S6_S29_L005       2409_S6_S47_L008 27_10_2016_6A_S30_L005       0406_S6_S30_L006 
             0.7822790              0.7170913              1.2903509              0.5022084 
      0508_S6_S43_L008        0511_S6_S9_L002 07_02_2017_6F_S10_L002       0705_S6_S18_L004 
             0.4575517              1.5178593              1.8893295              0.6809406 
      0810_S6_S12_L003 09_02_2017_6D_S29_L006 09_02_2017_6L_S42_L008        1410_S6_S1_L001 
             0.7387591              1.9627270              1.4965968              0.8661562 
      1609_S6_S16_L003  17_02_2017_6_S23_L005       1707_S6_S22_L004 19_10_2016_6B_S39_L007 
             0.7735276              1.5968769              0.5657576              2.2309920 
      2105_S6_S24_L005 23_02_2017_6C_S11_L003 23_02_2017_6F_S15_L003 27_10_2016_6B_S34_L006 
             0.8620617              1.3122498              0.9754176              1.7786902 
      3004_S6_S17_L004             3007S6_S29 
             0.4959165              1.2226773 
 

ADD REPLY
0
Entering edit mode

> idx.Blood.CD19 <- rowSums(counts(Blood.CD19.iterate, normalize = TRUE) >= 10) >=6
> Blood.CD19.keep <- Blood.CD19.coding[idx.Blood.CD19, ]
> nrow(Blood.CD19.keep)
[1] 12804
> Blood.CD19.keep  <- estimateSizeFactors(Blood.CD19.keep) 
using 'avgTxLength' from assays(dds), correcting for library size
> sizeFactors(Blood.CD19.keep)
NULL
> idx.Blood.CD19.all.rows <- rowSums(counts(Blood.CD19.iterate, normalize = TRUE) >= 10) >=26
> Blood.CD19.all.rows <- Blood.CD19.coding[idx.Blood.CD19.all.rows , ]
> nrow(Blood.CD19.all.rows)
[1] 6416
> Blood.CD19.all.rows  <- estimateSizeFactors(Blood.CD19.all.rows) 
using 'avgTxLength' from assays(dds), correcting for library size
> sizeFactors(Blood.CD19.all.rows)
NULL    

ADD REPLY
0
Entering edit mode

> sessionInfo()
R version 3.4.0 (2017-04-21)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows >= 8 x64 (build 9200)

Matrix products: default

locale:
[1] LC_COLLATE=English_United Kingdom.1252  LC_CTYPE=English_United Kingdom.1252   
[3] LC_MONETARY=English_United Kingdom.1252 LC_NUMERIC=C                           
[5] LC_TIME=English_United Kingdom.1252    

attached base packages:
 [1] grid      parallel  stats4    stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] sva_3.24.4                 BiocParallel_1.10.1        mgcv_1.8-22               
 [4] nlme_3.1-131               BiocInstaller_1.26.1       biomaRt_2.32.1            
 [7] Rgraphviz_2.20.0           plyr_1.8.4                 topGO_2.28.0              
[10] SparseM_1.77               GO.db_3.4.1                graph_1.54.0              
[13] genefilter_1.58.1          fdrtool_1.2.15             org.Hs.eg.db_3.4.1        
[16] pheatmap_1.0.8             hexbin_1.27.1              PoiClaClu_1.0.2           
[19] factoextra_1.0.5           RColorBrewer_1.1-2         geneplotter_1.54.0        
[22] annotate_1.54.0            XML_3.98-1.9               AnnotationDbi_1.38.2      
[25] Hmisc_4.0-3                ggplot2_2.2.1              Formula_1.2-2             
[28] survival_2.41-3            lattice_0.20-35            DESeq2_1.16.1             
[31] SummarizedExperiment_1.6.5 DelayedArray_0.2.7         matrixStats_0.52.2        
[34] Biobase_2.36.2             GenomicRanges_1.28.6       GenomeInfoDb_1.12.3       
[37] IRanges_2.10.5             S4Vectors_0.14.7           BiocGenerics_0.22.1       
[40] rjson_0.2.15               readr_1.1.1                tximport_1.4.0            

loaded via a namespace (and not attached):
 [1] bitops_1.0-6            bit64_0.9-7             tools_3.4.0             backports_1.1.1        
 [5] R6_2.2.2                rpart_4.1-11            DBI_0.7                 lazyeval_0.2.1         
 [9] colorspace_1.3-2        nnet_7.3-12             gridExtra_2.3           bit_1.1-12             
 

ADD REPLY
0
Entering edit mode

[13] compiler_3.4.0          htmlTable_1.9           scales_0.5.0            checkmate_1.8.5        
[17] stringr_1.2.0           digest_0.6.12           foreign_0.8-69          XVector_0.16.0         
[21] base64enc_0.1-3         pkgconfig_2.0.1         htmltools_0.3.6         limma_3.32.10          
[25] htmlwidgets_0.9         rlang_0.1.4             RSQLite_2.0             bindr_0.1              
[29] acepack_1.4.1           RCurl_1.95-4.8          magrittr_1.5            GenomeInfoDbData_0.99.0
[33] Matrix_1.2-12           Rcpp_0.12.13            munsell_0.4.3           stringi_1.1.6          
[37] zlibbioc_1.22.0         blob_1.1.0              ggrepel_0.7.0           splines_3.4.0          
[41] hms_0.3                 locfit_1.5-9.1          knitr_1.17              latticeExtra_0.6-28    
[45] data.table_1.10.4-3     gtable_0.2.0            assertthat_0.2.0        xtable_1.8-2           
[49] tibble_1.3.4            memoise_1.1.0           bindrcpp_0.2            cluster_2.0.6        

ADD REPLY
0
Entering edit mode
@mikelove
Last seen 1 day ago
United States

From the beginning, can you show an example of how the default size factor estimator produces NULL? Please show code and output, and print the sessionInfo(). This shouldn't happen, so if it is a bug, I'd want to fix it.

ADD COMMENT
0
Entering edit mode

hi,

Here the sizeFactors(dds) showing up as NULL is not a problem, because importing with tximport puts the sequencing depth normalization into normalizationFactors(dds). This takes precedence over sizeFactors(dds). You don't have to do anything extra (you don't have to use "iterate").

ADD REPLY

Login before adding your answer.

Traffic: 707 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6