4.2 HomeoRoq
HomeoRoq (Akama et al. 2014) estimates the probability that
homeolog expression ratios remain constant
between two experimental groups using RNA-Seq read count data for each homeolog.
Since this probability cannot be derived analytically,
HomeoRoq adopts a Bayesian framework
to approximate the posterior distribution of homeolog expression ratios
under the null hypothesis of no change between conditions.
The homeoroq() function provides a complete interface
for detecting homeologs whose expression ratios differ between two conditions.
By default, it performs 10,000 sampling iterations (iter_sampling = 1e4) per probability calculation
and repeats the calculation four times (chians = 4).
However, for robust and reproducible estimation of statistical significance (i.e., p-values),
it is recommended to run multiple independent sampling replicates,
each with a sufficiently large number of iterations.
To reproduce the settings used in the original HomeoRoq implementation (Akama et al. 2014),
set iter_sampling = 1e4 and chains = 10.
4.2.1 Inputs
The homeoroq() function requires an input object of class ExpMX,
same to the hobit() function.
4.2.2 Test
Once an ExpMX object is prepared,
normalize the counts if the data have not already been normalized,
then run the test using the homeoroq() function.
4.2.3 Outputs
The result (x_output) is a data frame with one row per homeolog,
summarizing the statistical test results.
## gene pvalue qvalue sumexp__wet__C_hirsuta sumexp__wet__C_amara sumexp__dry__C_hirsuta sumexp__dry__C_amara ratio__wet__C_hirsuta ratio__dry__C_hirsuta ratio_sd
## 1 CARHR000190_H 0.241925 1 383.26142 475.04089 419.43533 285.21986 0.4465343 0.4689173 0.11000959
## 2 CARHR000660_H 0.942875 1 14.57256 46.19000 15.95469 40.22082 0.2398279 0.2567345 0.04073867
## 3 CARHR000770_H 0.990150 1 385.98379 44.96460 850.62370 95.35341 0.8956613 0.9497932 0.01737532
## 4 CARHR000890_H 0.520575 1 181.89090 107.32037 12.38467 10.08555 0.6289205 0.1034599 0.12263357
## 5 CARHR001940_H 0.978275 1 110.98129 81.00309 318.05281 285.09804 0.5780746 0.7970132 0.05006071
## 6 CARHR003740_H 0.197475 1 1549.17077 2270.28481 1179.77571 2596.99673 0.4055999 0.3419580 0.05277229gene: Homeolog ID. The gene names defined in the first column of the mapping table.pvalue: p-value from the likelihood ratio test.qvalue: Adjusted p-value using the Benjamini-Hochberg method.sumexp__*__$: Total read counts for subgenome$under condition*.ratio__*__$: Homeolog expression ratios for subgenome$under condition*.ratio_sd: Standard deviation of homeolog expression ratios calculated from observed counts.
The total read counts (sumexp__*__$) and homeolog expression ratios (ratio__*__$)
are computed directly from the observed read counts.
Note that this differs from the hobit() function,
which calculates its statistics from MCMC samples.