Illumina to contribute methods to Open Source

[skruglyak]

Is this https://github.com/Sequencing now considered outdated? Or will content be transfered to 'Illumina'?

What are the major changes from isaac 0.1.x to isaac2?

Thanks for your interest. We will transfer content for the primary tools that are switching licenses and will continue to be developed and supported to /Illumina.

For Isaac2, here are the main changes:

• Split read support to facilitate SV detection (similar to BWA-MEM)

• Improved detection of ambiguous reference sequences (e.g. long repeats and pseudogenes)

• Improved indel recall and precision when using this version with the downstream caller (Starling)

• Support for BAM as input

• Various improvements in IO management and memory usage

• Bug fixes

[sklages]

Is this https://github.com/Sequencing now considered outdated? Or will content be transfered to 'Illumina'?

What are the major changes from isaac 0.1.x to isaac2?

[skruglyak]

Just the kind of information I was asking for. Thanks!

Can minimal/optimal hardware specs be noted on the respective pages so people can make an informed decision on whether they should try using that package. e.g. isaac (and possibly isaac2, going to try it soon) needs reasonably potent hardware.

Yes, I will ask each developer to post some info on hardware requirements on GitHub. Isaac tends to be most resource intensive. We typically run it on a node with 32 CPUs and 128 GB of RAM to align a 30X genome in about 2 hours.

[GenoMax]

Just the kind of information I was asking for. Thanks!

Can minimal/optimal hardware specs be noted on the respective pages so people can make an informed decision on whether they should try using that package. e.g. isaac (and possibly isaac2, going to try it soon) needs reasonably potent hardware.

[skruglyak]

Can I make a couple of suggestions? It would be helpful if developers could also add notes about what OS (and library versions) the code has been successfully compiled on to the GitHub pages. If statically linked binaries could be made available (in addition to the source) that would save novice users a lot of heartache (and may actually help increase usage of these tools).

Thank you for the excellent suggestions. For OS and library versions, this is the plan. You can see the excerpt from the Manta posting below. For binaries we have to gauge the level of interest as well as scope the effort required. We are experimenting with the idea and you can take a look here for an example with Manta.

Thanks,
Semyon


"Manta requires a compiler supporting most of the C++11 standard. These are the current minimum versions enforced by the build system:

python 2.4+
gcc 4.7+ OR clang 3.1+ (OR Visual Studio 2013+, see windev note below)
libz (including headers)
Runtime prerequisites

python 2.4+
Operating System Guidelines

Linux

Manta is known to build and run on the following linux distributions (with additional packages as described below):

Ubuntu 12.04,14.04
CentOS 5,6,7
OS X

Manta builds and passes basic tests on OS X 10.9, but full WGS analyses are not tested for this platform.

Windows

Manta does not build or run on windows. Library-level compilation is possible for Visual Studio users. See Contributor section below."

[GenoMax]

Hi GenoMax,

We will not be providing support of these components through the official customer support channels (those are dedicated to commercial releases). However, each lead developer will set up something like a Google board to answer questions. Also, we will be providing documentation with each release.

thanks,

Semyon

Can I make a couple of suggestions? It would be helpful if developers could also add notes about what OS (and library versions) the code has been successfully compiled on to the GitHub pages. If statically linked binaries could be made available (in addition to the source) that would save novice users a lot of heartache (and may actually help increase usage of these tools).

[skruglyak]

@Semyon: As I remember, Illumina's policy was that code released on GitHub was provided "as is" and was not supported. Is that policy changing?

Is GitHub "issues" section the best place to report problems with code already on GitHub?

Hi GenoMax,

We will not be providing support of these components through the official customer support channels (those are dedicated to commercial releases). However, each lead developer will set up something like a Google board to answer questions. Also, we will be providing documentation with each release.

thanks,

Semyon

[GenoMax]

@Semyon: As I remember, Illumina's policy was that code released on GitHub was provided "as is" and was not supported. Is that policy changing?

Is GitHub "issues" section the best place to report problems with code already on GitHub?

[HESmith]

Hi Semyon,

Thanks for the quick response, and it's good to know that Isaac2 is not limited to human. You may want to share that information with your colleagues in tech support, since they were not able to provide a clear answer (they also told me that original Isaac was human-specific, although that may have been in the context of BaseSpace).

-Harold

[skruglyak]

Hi Semyon,

Quick question: is Isaac2 limited to human genomes (as was the original Isaac), or are non-human references useable?

Thanks,
Harold

Hi Harold,

Thanks for your interest. Neither version of Isaac is limited to human. (Though most of our testing is done on human) If you want to run on another organism, you would need to generate the index. We typically use a 128 GB RAM node and the process takes half a day. Of course you only need to do it once. But once you have the index, the method applies to any organism. In the next version of Isaac, you will not need to generate an index.

thanks,
Semyon

[dgaston]

Excellent news. Many of us are interested in testing out elements of workflows and tools without feeling locked into the "click-and-run" aspects of some of the more built-in tools targeted at non bioinformaticians.

[HESmith]

Hi Semyon,

Quick question: is Isaac2 limited to human genomes (as was the original Isaac), or are non-human references useable?

Thanks,
Harold

[SNPsaurus]

Thanks! Nice to see this sort of thing happening.

[skruglyak]

Hi everyone,

Illumina will release secondary analysis tools using either the GPLv3 or BSD 2-Clause license. The tools can be found at https://github.com/illumina. We have started with the Isaac2 aligner (an improvement on the previously published version) and the Manta structural variant caller (manuscript submitted). In the coming weeks, we will add the Strelka somatic variant caller, the Canvas copy number caller (manuscript in preparation), and various other tools.

I will be happy to answer any questions that you may have.

Thank you,

Semyon
(member of the Bioinfo group at Illumina)