[BioC] Rsubread : featureCounts PEReadsReordering=TRUE deletes unpaired reads?

[Ryan C. Thompson]

Hello,

I am using the featureCounts function in Rsubread to count RNA-seq 
reads. Since my BAM files are coordinate-sorted paired-end, I am using 
"isPairedEnd=TRUE" and "PEReadsReordering=TRUE". When I run my script, I 
see the following in the output:

Process /gpfs/home/rcthomps/Projects/cyno/tophat_results/R77_CN7314_P4 ...
    Resort the input file ...
    2718714 unpaired reads were ignored in reordering.
    Total number of fragments is : 1653582
    Number of successfully assigned fragments is : 1529491 (92.5%)
    Running time : 3.39 minutes

In particular, I am worried about the line "2718714 unpaired reads were 
ignored in reordering." The documentation for PEReadsReordering doesn't 
say anything about discarding unpaired reads, so I'm wondering whether 
this is intended behavior, and whether there is a way around this? If I 
sort my bam files by name so that pairs are adjacent in the file, will 
subread still discard the unpaired ones, or will it count all of them?

-Ryan Thompson