sasi-arunachalam/mirnaseq_rnw

miRNAseq regulatory networks

miRNAseq_rnw

miRNAseq regulatory networks

1. expression matrix mature miRNA (miRBase_mysql.txt):

dim(x) [1] 2588 752

2. formating matrix (format_miR_matrix.R): matrix format: rownames: GENE/miR ID, colnames: patient code: XXXX.01(case)/11(control)

miRNAs_752_maduros.txt

3. filter raw counts: filter targets with a minimum of 5 counts in at least 25% of the samples(752) = 188 low_counts_filter_miRNA.R

dim(miRNA) [1] 715 752 miRNAs_752_maduros_filtro.txt

4. density plot

density_plot.miR_RNAseq.R

5. Normalization miRNA_normalization.R "Adjusting the data by cpm or total count scaling introduces more variability to the data, whereas all other methods resulted in more similar distribution across all samples." " The distributions are more similar and centered at zero when the data are normalized using UQ, TMM, DESeq, cyclic loess and quantile normalizations" "UQ and TMM decreased the variance across all miRNAs compared with the raw data" R statistical environment (v3.1.2) Bioconductor (3.0) Cyclic loess and quantile normalization were performed using the normalizeBetweenArrays() function in the limma package (v3.22.4) TMM and UQ normalizations were performed using calcNormFactors() in the edgeR (v3.8.5) package the function estimateSizeFactors() in the DESeq (v1.18.0) package was used to normalize the count data using size factors (http://bib.oxfordjournals.org/content/early/2015/04/17/bib.bbv019.full.pdf+html)

miRNA_752_normTMM.txt TMM1.miRNA.753.pdf raw.miRNA.753.pdf

6. add RNAseq expression data to miRNAseq for mi miRNA-miRNA miRNA-mRNA interactions expression_matrix_miRNAs.R
7. add RNAseq network mRNA-mRNA interaction for whole network

red.RNAseq.final.752.txt = 99.987% dpi 0.1 11202 interactions (with no header to cat under miRNA network) red.MIRseq.final.752.txt = 99.980% dpi 0.1 cat red.MIRseq.final.752.txt red.RNAseq.final.752.txt > red.final.752.txt

red.final.752.txt

for 86 samples matrices enfermos <- read.table("miRNAs_enfermos_maduros.txt", sep="\t", header=T, row.names=1) sanos <- read.table("miRNAs_sanos_maduros.txt", sep="\t", header=T, row.names=1)

1. expression matrix mature miRNA (miRBase_mysql.txt):

dim(enfermos) [1] 2588 752 dim(sanos) [1] 2588 752

2. formating matrix (format_miR_matrix.R): matrix format: rownames: GENE/miR ID, colnames: patient code: XXXX.01(case)/11(control)

miRNAs_enfermos_maduros.txt miRNAs_sanos_maduros.txt

3. filter raw counts: filter targets with a minimum of 5 counts in at least 25% of the samples(86) = 22 low_counts_filter_miRNA.R

dim(enfermos)
[1] 690 86 dim(sanos) [1] 658 86 miRNAs_enfermos_maduros_filtro.txt miRNAs_sanos_maduros_filtro.txt set two groups to intersection (633) miRNAs_752_maduros_filtro.txt

4. density plot

density_plot.miR_RNAseq.R

5. Normalization miRNA_normalization.R "Adjusting the data by cpm or total count scaling introduces more variability to the data, whereas all other methods resulted in more similar distribution across all samples." " The distributions are more similar and centered at zero when the data are normalized using UQ, TMM, DESeq, cyclic loess and quantile normalizations" "UQ and TMM decreased the variance across all miRNAs compared with the raw data" R statistical environment (v3.1.2) Bioconductor (3.0) Cyclic loess and quantile normalization were performed using the normalizeBetweenArrays() function in the limma package (v3.22.4) TMM and UQ normalizations were performed using calcNormFactors() in the edgeR (v3.8.5) package the function estimateSizeFactors() in the DESeq (v1.18.0) package was used to normalize the count data using size factors (http://bib.oxfordjournals.org/content/early/2015/04/17/bib.bbv019.full.pdf+html)

miRNAs_86_maduros_norm.txt miRNAs_86_maduros_norm_enfermos.txt miRNAs_86_maduros_norm_sanos.txt TMM.miRNA.86.pdf raw.miRNA.86.pdf

6. add RNAseq expression data to miRNAseq expression_matrix_miRNAs.R
7. add RNAseq network mRNA-mRNA interaction for whole network

enfermos.99.987.dpi.0.1.sif.txt -> red.enfermos.RNAseq.86.txt = 99.987% dpi 0.1 14280 interactions (with no header to cat under miRNA network) sanos.99.987.dpi.0.1.sif.txt > red.sanos.RNAseq.86.txt = 99.987% dpi 0.1 14890 interactions sanos.mir.99.980.dpi.sif.txt > red.sanos.MIR.86.txt = 99.980% dpi 0.1 24401 interactions cat red.sanos.MIR.86.txt red.sanos.RNAseq.86.txt > red.final.sanos.txt enfermos.mir.99.980.dpi.sif.txt > red.enfermos.MIR.86.txt = 99.980% dpi 0.1 23582 interactions cat red.enfermos.MIR.86.txt red.enfermos.RNAseq.86.txt > red.final.enfermos.txt

probes

• miR752 -> probes_miRNA.txt yes 0 | head -n 715 > 1.txt paste probes_miRNA.txt 1.txt > clave_MIR_752.txt
• RNAseq752 -> probes_RNAseq_752 yes 1 | head -n 15972 > 1.txt paste probes_RNAseq_752 1.txt > clave_RNA_752.txt
• miR86 -> probes_86.txt yes 0 | head -n 633 > 1.txt paste probes_86.txt 1.txt > clave_MIR_86.txt *RNAseq86 -> probes_RNAseq_86 yes 1 | head -n 15136 > 1.txt paste probes_RNAseq_86 1.txt > clave_RNA_86.txt

cat clave_MIR_86.txt RNAseq_86_genes.txt RNAseq_86_TF.txt > clave.final.86.txt cat clave_MIR_752.txt RNAseq_752_genes.txt RNAseq_752_TF.txt > clave.final.752.txt

enfermos <- read.table(file = "enfermos.86.adj.txt", header = T, sep = '\t') p <- enfermos p[p < 0.16482207] <- 0 sum(enfermos != 0)/2 sum(p != 0)/2