A working command line for Illumina's iSAAC ?

[oiiio]

I just got hold of the iSAAC aligner from Illumina, but it seems to use a very different type of command line than most of the other aligners I have used, particularly when specifying a FASTQ. Does anyone have a command line that has been working for them?

[sklages]

What problems occured? Did you read the output of

./bin/isaac-align --help

[oiiio]

So I realized that the FASTQ names must be specifically named as lane1_read1.fastq lane1_read2.fastq

Now just trying to index the hg19 reference with isaac-sort-reference . Seems to be taking quite a long time

[GenoMax]

Now just trying to index the hg19 reference with isaac-sort-reference . Seems to be taking quite a long time

Or you could download it from Illumina. It is available along with the HiSeq Analysis Software.

[oiiio]

Thanks I'm downloading it now, although I wonder if this download will finish before my indexing command...

Even though it looks like the downloadable ref is based on the same UCSC hg19 I am using, I can't help but feel a little loss of experimental control when it comes to comparisons against the other software I ran

[oiiio]

After downloading the reference I'm now hitting this error:


2013-06-10 17:52:33 [0x11d36a0] __gnu_parallel:sort for 64000000 elements
terminate called recursively
terminate called recursively
./isaac-run.sh: line 5: 40176 Aborted (core dumped) isaac-align --reference-genome IsaacIndex/sorted-reference.xml -b isaac/ --base-calls-format fastq -m 12 --verbosity 3

Has anyone put a successful run through with Isaac yet?

[craczy]

Hi oiiio

You would need a lot more than 12GB memory. Try "-m 48" instead.

Thanks,

Come

[GenoMax]

Has anyone managed to get "isaac-align" to work stand-alone on the command line and if so on what OS?

"Isaac-align"er is one of the components of HiSeq Analysis Software (HAS) and I have managed to get it to work when run as a part of a HAS run (but only under SGE, no luck with LSF).

Trying to get it to work stand-alone has been an exercise (Oiiio: I could not get it to work with the options you noted, on individual files etc). I thought I had managed to to it to work using the following command line (on a rocks v. 5.4.2 cluster)

isaac-align -b /Flow_cell_folder/Data/Intensities/BaseCalls -r /path_to/ISAAC_genomes/hg19/Homo_sapiens/UCSC/hg19/Sequence/IsaacIndex/sorted-reference.xml -o /output_dir -m 48

It created binary demultiplexed files (*.dat) (ran for about 8 hours, this was a big flowcell) before it died with following error

Error: boost::exception: Throw location unknown (consider using BOOST_THROW_EXCEPTION)
Dynamic exception type: boost::exception_detail::clone_impl<boost::exception_detail::current_exception_std_exception_wrapper<std::bad_alloc> >
std::exception::what: std::bad_alloc
[boost::tag_original_exception_type*] = St9bad_alloc

Illumina tech support says that the aligner part from HAS is not meant to be run as a standalone program and they pointed me to an open source project (https://github.com/sequencing/isaac_aligner) where the aligner from HAS has been released as a "developer" version i.e. they provide no support. There is no documentation (and as far as I can tell, not all of the command line options seem to work) in the github package.

[craczy]

Hi GenoMax,

if you managed to run it under HAS, why don't you use the command line generated by HAS? That command line can be found either on the standard output, or at the beginning of the log file "Logging/Isaac.stderr.txt" that you will find in the HAS output directory.

Regarding the specific behaviour that you observed when trying to run directly from the command line, there is clearly an issue with the memory settings, but it is difficult to diagnose without more information. Could you provide the output of:

ulimit -a
tail -n 9999 /proc/sys/vm/*
cat /proc/meminfo

and also the last couple hundred lines of stderr before the crash.
Thanks,

Come

[GenoMax]

Hi GenoMax,

if you managed to run it under HAS, why don't you use the command line generated by HAS? That command line can be found either on the standard output, or at the beginning of the log file "Logging/Isaac.stderr.txt" that you will find in the HAS output directory.

Come

I could do that (I may give it a try) but the point was one should be able to use the aligner without having to worry about all other parts of HAS and options that need to go with it. I understand from our local FAS that isaac was developed with one aim and that was to finish alignments in the shortest time possible. I want to be able to point "isaac-align" to a set of sequence files (should not matter which flowcells they came from), an ISAAC reference genome and a few basic options to get aligned bam files.

If that is not possible (or is not in the scope of isaac package) then it would be best to have that spelled out clearly.

I am not sure if you are @Illumina/UK (part of the isaac development team?). If you are, then we could take this conversation temporarily offline to omit thread-clogging info that may be specific to problem I am noticing.

[LVAndrews]

Has anyone had success with command-line isaac? I have a non-model organism I would like to push through this pipeline (ponderosa pine). If I make success, will try to post some commands here for future reference.

[GenoMax]

I will have to look through my results from back last year. I have since used HAS several time though only for human data.

Before you start be aware that ISAAC/HAS was designed to take over an entire server (using *all* hardware resources on it). It is not going to run on a cluster.

[oiiio]

We had the command on this GCAT page work:

http://www.bioplanet.com/gcat/reports/530/variant-calls/illumina-100bp-pe-exome-150x/isaac-isaac/compare-156-adsqyilqlu-571/group-read-depth

[sklages]

Is anyone running current git version? It still seems to be a problem running isaac-aligner on its own.
Currently it is always stuck here with memory issues (well, not necessarily the amount, I've tried to run on 512GB machines, nor limits).
isaac part of HAS is quite old.

[GenoMax]

@Sven: Are you getting an error or does it seem to just hang?

I have not successfully run isaac-aligner as a standalone app in the past (going to upgrade to the latest version from git now, thanks for the heads-up) since there was no support available.