# 4. (DIFFICULT): Create a text-tabulated file with raw counts. Tip: assays(dds)$counts

rawCounts <- assays(dds)$counts
# ALTERNATIVE:
rawCounts <- counts(dds, normalized=FALSE)

write.table(data.frame(ID=row.names(dds),rawCounts), file="my_raw_counts.tsv", sep="\t", row.names=FALSE)

#
# 5. (DIFFICULT): Create a text-tabulated file with normalized counts. Tip: assays(dds)$mu
#           NOTE: It is illustrative to look at normalized counts of differentially
#               expressed genes with good pvalues.

normCounts <- t(t(assays(dds)$counts)/sizeFactors(dds))
# ALTERNATIVE:
normCounts <- counts(dds, normalized=TRUE)

write.table(data.frame(ID=row.names(dds),normCounts), file="my_norm_counts.tsv", sep="\t", row.names=FALSE)

# EXPLANATION (for experienced R users):
# the "t()" function (transpose), swap rows by columns in a matrix. So:
# t(assays(dds)$counts)/sizeFactors(dds) divides each former column (now, each row) by the corresponding sizeFactor.
# Finally, by appying again t() to the result, we obtain the original row and column dimensions.
# 


#
# 6. (DIFFICULT): add columns with normalized counts to the FINAL table (create file).
# (using "normCounts" calculated in the task #5)

FINAL_P5C50vsP5C004_withNormCounts <- data.frame(
                                     GENEID = row.names(P5C50vsP5C004),
                               log2BaseMean = log2(P5C50vsP5C004$baseMean),
                                  log2Ratio = P5C50vsP5C004$log2FoldChange,
                           STDERR_log2Ratio = P5C50vsP5C004$lfcSE,
                                     pvalue = P5C50vsP5C004$pvalue,
                                    padjust = P5C50vsP5C004$padj,
                                    normCounts
                                     )

write.table(FINAL_P5C50vsP5C004_withNormCounts, file="RESULTS_P5C50vsP5C004_withNormCounts.tsv", sep="\t", row.names=FALSE)