library(DESeq2)
deg_dog_df <- results(DESeq(DESeqDataSetFromMatrix(countData=rsc_exp_dog,
colData=data.frame(condition=factor(col_data_dog$condition,
levels=c("normal","cancer")),
row.names=rownames(col_data_dog)),
design=~condition)))>head(deg_dog_df)
log2 fold change (MAP): condition cancer vs normal
Wald test p-value: condition cancer vs normal
DataFrame with 6 rows and 6 columns
baseMean log2FoldChange lfcSE stat pvalue padj
<numeric> <numeric> <numeric> <numeric> <numeric> <numeric>
ENSCAFG00000014413 17.3 -0.4098 0.613 -0.6684 0.5039 0.655
ENSCAFG00000014412 117.8 0.6927 0.399 1.7362 0.0825 0.185
ENSCAFG00000014410 45.4 -0.4414 0.538 -0.8211 0.4116 0.575
ENSCAFG00000014416 392.7 0.1547 0.489 0.3162 0.7518 0.846
ENSCAFG00000014415 65.3 -0.3228 0.314 -1.0291 0.3034 0.468
ENSCAFG00000020948 513.7 0.0463 0.738 0.0628 0.9499 0.969
>dim(deg_dog_df)
[1] 13572 6
In the above output table, "padj" (i.e.,padj) is thep-value
that has been adjusted for multiple hypothesis tests (i.e., a
q-value) using the Benjamini-Hochberg correction [32]. The
"DESeqDataSetFromMatrix" function constructs an object
containing the sample information and mRNA-seq count data
for each (expressed) gene; the "DESeq" function performs· ··
·
····
·· ··
···················−5 ··05−10 −5 0 5 10
X1X2species
dog
human
condition
·
·cancer
normal·Fig. 2Principal coordinates analysis (PCoA) showing combined transcriptome
profiling data sets from human bladder samples and dog bladder samples. Even
in this unbiased (i.e., blinded to sample group type) analysis of the data, there is
a similar line of separation between normal and cancer samples in both dogs
and humans, indicating a high degree of similarity between the human and dog
bladder cancer transcriptomes at the levels of gene functions302 Stephen A. Ramsey