Rsubread package not available? Warning resolved
2
0
Entering edit mode
NDowell ▴ 20
@ndowell-4997
Last seen 8.0 years ago
0
Entering edit mode
Wei Shi ★ 3.5k
@wei-shi-2183
Last seen 1 day ago
Australia/Melbourne/Olivia Newton-John …

Dear Noah,

We are still thinking about whether we should make Rsubread available on Mac or Windows laptops. I can certainly see the advantage of doing this. However, this will require considerable amount of work for developing and maintaining it because many functions in the package are written in C language and we are going to include more C functions into the package in the future. Moreover, the read alignment will consume most of the computational resources of a laptop and I do not think you can use your laptop for other work during the period of read mapping. So it might be best to run the read alignment on a server/supercomputer, rather on a laptop.

Cheers,
Wei

0
Entering edit mode

Dear Wei,

i think you might consider the following three reasons why it would be interesting to have Rsubread available in Mac OS X systems:

1. i've used Rsubread to teach my students about mapping reads without leaving the R console by using a small sample of reads (e.g. 1e5, 1e6), if eventually one makes a teaching course where students come with laptops, a significant fraction of them come now with mac. this could apply to all the BioC courses that take place throughout the world :)

2. apple not only sells laptops but also high-performance computing solutions. those users would also benefit of having Rsubread running in Mac OS X.

3. Mac OS X runs on top of a unix system, so it should not be all too difficult to have and maintain ANSI C code running in both, a linux and a Mac OS X system.

cheers, robert.

0
Entering edit mode

Dear Wei,

Thanks for your response. I was going to use Rsubread with a small(ish) data set on a desktop with a little more memory than the standard laptop. I like the approach behind Rsubread and I wanted to do some comparisons with other aligners on small data sets before making a choice.

Of course I understand as a maintainer you have to make some careful choices where and what to put your effort into.

I should have a server up and running soon (at my new institute; not UCLA) and will consider using Rsubread at that time.

Best, Noah

0
Entering edit mode

FWIW, I've had pretty good results with some older 36-cycle Illumina RNAseq data using Rsubread, just aligning it on my laptop (granted it's a dual-core i7 with 16GB of RAM and 2 SSDs). For some of the RNAseq data from CD34+CD38- cells and neutrophils that Andrew Smith was kind enough to send, it took about 30 minutes, and I was able to do other things at the same time. Again, that might have something to do with my using 7.4GB for the index, having another 8.6GB for other processes, and running Linux. But mostly the RAM, I think.

YMMV... Rsubread has some terrific features, I am torn between using it all the time, using BowTie/TopHat/Cufflinks/cummeRbund all the time, or both. Last time I just used both, but with our RNAseq/QC pipeline, if I want to re-align against hg18 I either have to use bedtools and liftOver, re- run the pipeline, or use Rsubread. The latter is more convenient, even if the former does have some more tools for removing PCR dupes, looking at alternative splicing, etc.

Either one runs fine on our servers, of course, and both go a lot faster with 24 cores and 48GB of RAM :-)

0
Entering edit mode

One other thing Noah, do you have Xcode and the R tools for Mac OSX installed? I'm using my wife's Macbook Air at this particular instant, and I'm tempted to try compiling Rsubread, but I don't really have time to debug something for fun right now. Still -- it would be interesting to see if that resolved your issues. It's not like laptops are being produced with less RAM and smaller processors as time goes by, and there are old-but-still-very-useful RNAseq datasets out there. And Rsubread is FAST so it's by far the most sensible choice for a laptop.

1
Entering edit mode

Dear Tim, Robert and Noah,

I think we will eventually make Rsubread available on Mac OS X system (but may not on Windows). Hopefully it will become available on the next bioc release. But Yang Liao, who wrote the entire C program for the read alignment, has to submit his Ph.D thesis February next year, and also we are now busy with publishing Rsubread. But I'm quite sure it should be available sometime next year.

Rsubread uses the spinlock library to enable multithreaded running of the read alignment, which is not supported on Mac. We'll have to change it to the mutex library which is supported by both Mac and Linux for portability (however, we found spinlock is more efficient than mutex:-( ).

The align() function in Rsubread can map junction reads from the RNA- seq data in addition to the exonic reads, although it won't give you the information of where the junction locations are (it gives the mapping location of the junction read in one of the exons it spans). The subjunc() function in Rsubread package can however be used to find the exact exon junction locations in the reference genome.

Cheers,
Wei

0
Entering edit mode

TIm,

Thanks for providing your experiences on using different aligners. I used R/Bioconductor for tiling and expression arrays as a grad student but now I am using next generation sequencing as a postdoc so I am getting up to speed with some new packages like Rsubread.

Yes, I have the Xcode developer tools installed and have installed from source recently for other packages with no errors.

Best, Noah

0
Entering edit mode

Hi all,

On Thu, Dec 8, 2011 at 12:10 PM, Noah Dowell <noahd at="" ucla.edu=""> wrote:
> TIm,
>
> Thanks for providing your experiences on using different aligners.
I used R/Bioconductor for tiling and expression arrays as a grad
student but now I am using next generation sequencing as a postdoc so
I am getting up to speed with some new packages like Rsubread.
>
> Yes, I have the Xcode developer tools installed and have installed
from source recently for other packages with no errors.


Wei can correct me if I am wrong, but I believe it's not as simple as building Rsubread from source on Mac (or Windows). If it were, Bioconductor would provide a binary package on those platforms.

I believe the problem is that certain constructs (e.g. phtread) are used in such a way that is not portable to Mac OS X or Windows,

Dan

0
Entering edit mode

Hi Dan,

Yes, that's right. pthread is a construct which is not portable between linux and Mac. But there might be other such constructs as well. We will need to have a close look at them.

Thanks, Wei

0
Entering edit mode
@gordon-smyth
Last seen 19 minutes ago
WEHI, Melbourne, Australia

29 August 2019

Starting from version 1.35.24, Rsubread is now available for Windows as well as the other platforms. For the moment, this is only for the developmental version of Bioconductor but it will migrate to the release version soon.