Strandedness of dual-RNAseq samples

[_Ramya]

Hi,

I am processing data from a dual-RNAseq experiment where one sample contains reads from both an insect host and a bacterial symbiont. I seperate reads belonging to bacteria or the insect host by mapping the sequencing data against the respective references. After aligning, I did not know how the libraries were prepared for sequencing so I used infer_experiments to determine the strandedness before proceeding to get read counts.

Reads were first aligned to the bacterial reference genome. And infer experiments result was the following for a sample_X

Results for sample_X (reads aligned to bacterial ref):
This is PairEnd Data
Fraction of reads failed to determine: 0.0018
Fraction of reads explained by "1++,1--,2+-,2-+": 0.0881
Fraction of reads explained by "1+-,1-+,2++,2--": 0.9101

I concluded the samples are stranded (htseq: reverse, featurecounts 2) libraries.

I do not yet have the aligment for the same sample_X against the host reference. But I wanted to find strandedness in another sample_Y with reads that aligned to the insect host reference genome:

Results for Sample_Y (reads aligned to host ref):
This is PairEnd Data
Fraction of reads failed to determine: 0.0000
Fraction of reads explained by "1++,1--,2+-,2-+": 0.2071
Fraction of reads explained by "1+-,1-+,2++,2--": 0.7929

When I increased the number of reads subsampled (2,000,000):

This is PairEnd Data
Fraction of reads failed to determine: 0.0000
Fraction of reads explained by "1++,1--,2+-,2-+": 0.3701
Fraction of reads explained by "1+-,1-+,2++,2--": 0.6299

The results testing strandedness on host aligned reads are less clear to me. I want to believe these are also stranded (htseq reverse, featurecounts 2) because the reads come from the same source. Am I wrong?

This is probably silly, but is it possible that in dual-RNAseq data the strandedness of the host and bacterial symbiont reads are different?

Also, the reference genome used for the host alignment is the predicted coding sequences only. Is this in any way a problem for the tool to predict strandedness?

I am new to this forum and to rnaseq analysis -any input is appreciated! Thanks!