•   
  •  
  •  
  • Login
  • |
  • Register
RSS Feed Print
Open source versus commercial software
HARSHA RAJASIMHA
Posted: Tuesday, August 16, 2011 12:08 PM
Joined: 3/4/2011
Posts: 11


Given that NGS data analysis is rapidly evolving and new algorithms are still being published, do you prefer running / trying open source data analysis tools manually? OR would you rather use a start-to-finish commercial solution that may not have the latest algorithms (or their versions) implemented yet?
AzcoBiotech
Posted: Friday, August 26, 2011 11:12 AM
Joined: 7/26/2011
Posts: 1


Although "Open Source" is good for developers, I personally like Open Source, it is not good for most users.  Users have a job to do, typically solve some biologic problem, and most open source software is buggy and prone to problems which hinders them from being able to solve their problems.   Commercial software alleviates these problems and "cans" the application so that researchers can get their job done without having to modify code or worry about how the application works...it just works and they solve problems.
Eduardo Pareja
Posted: Saturday, August 27, 2011 2:36 AM
Joined: 2/16/2011
Posts: 4


I think that you are right in your point that NGS data, technologies and algorithms are so rapidly evolving that, for me,  the usual time required to release to the market a commercial software and new versions is almost impossible. In addition, the best solutions and implementations for NGS data analysis are licensed in a way that many times can not be incorporated in commercial, closed, packages.  

Another point is that for many centers, research institutions and hospitals, is a key factor to be able to integrate the bioinformatics solution with their own Information Systems and it requires, the majority of the times, to be able to access to the code and modify it a little.

With regard to "start-to-finish" solutions, I think that the only possibility to have a real start-to-finish solution is implement pipelines or workflows based in different open source modules that fit exactly with your needs and obviously this is almost impossible with proprietary, closed, and not flexible solutions.


Eduardo Pareja
Posted: Saturday, August 27, 2011 2:42 AM
Joined: 2/16/2011
Posts: 4


With regard to the fact that the open source is buggy, ALL the software is buggy but the difference is that in an open source project you have the possibility to fix a bug and you are not obligated to wait for the next version and be lucky that this particular bug is fixed.

 Still, to may knowledge, a lot of users run every day a lot of open source solutions without altering a single line in the code.


Tom Schwei
Posted: Monday, August 29, 2011 11:49 AM
Joined: 1/26/2011
Posts: 3


I'll say at the outset that I represent DNASTAR, a commercial software provider.  So, clearly, I am biased in my position on this issue.  I believe there can be a role for both open source software and commercial solutions.  Most open source solutions require a bit of bioinformatics capability.  If a person has that capability or access to it and the solution works, I say go for it. 

 

The reasons I most often hear from people interested in our commercial software solutions for why they want to try them are:

  1. They aren't bioinformaticians and don't have easy access to one (or can't afford one).
  2. They don't want to wait in line at a core bioinformatics facility for analysis of their work, some times having to do this multiple times if iterative analysis is required, which is often the case.
  3. They aren't comfortable with command line programs.

A summary of why we hear many people prefer commercial software solutions is that they know their work best and what results make sense and don't make sense.  If they can have affordable, easy to use tools available to them to do their own analysis, that is the best solution.

 


Daniel MacArthur
Posted: Tuesday, August 30, 2011 5:30 AM
Joined: 12/1/2010
Posts: 1


I think most users at large genomics facilities prefer to create roll-their-own pipelines from open-source or in-house code - this is typically the best solution when you have a very good informatics team, lots of data, and specialised analytical demands that don't necessarily fit into the standard model addressed by commercial packages.

However, for smaller labs with less informatics support, buying an off-the-shelf commercial solution can end up being far more cost-effective than having researchers spend months of their time learning how to code and building their own buggy pipelines. Of course, the benefits of this approach depend on having a commercial provider willing to add in the latest analytical advances as they become available.


HARSHA RAJASIMHA
Posted: Wednesday, August 31, 2011 10:47 AM
Joined: 3/4/2011
Posts: 11


Great comments and thoughts on this dilemma that most initial investigators are facing - whether to invest in a COTS application or build their own pipelines. How do you see this dilemma sort itself in the next 3-5 years? What do you imagine / expect to be in common practice?
Tom Schwei
Posted: Thursday, September 1, 2011 8:55 AM
Joined: 1/26/2011
Posts: 3


I think we'll actually see a shift to more commercial software, whether provided by instrument providers, commercial software companies or combinations of the two.  The open source software met a need early in the next-gen revolution, when commercial software companies didn't yet have algorithms to deal with the new issues associated with this new type of data.  That's not uncommon when a new technology arises.  However, as the next-gen technologies mature, the instrument companies and third party software providers can evolve their tools and "catch up".  In addition, third party software providers put more emphasis on ease of use, which is more important to people who join in as technologies mature.  Early adopters are more inclined to use anything that works.  As next-gen sequencing technologies continue to move into more and more labs, especially with the benchtop units now available, more non-bioinformatic life scientists will have access to the technology and need affordable, easy to use tools at their fingertips. 

 

There will still be those (large genome centers, for example) who prefer to roll their own, but I think we'll see a shift to more commercial third party software in the future.


Peter Grant
Posted: Friday, September 9, 2011 2:56 PM
Joined: 7/26/2011
Posts: 1


As a representative for a "commercial" software solution, I'd like to say that everyone seems to want to put each potential solution...open source or commercial...into separate boxes which have no interaction at all. Admittedly, most commercial solutions are more or less a black box and do not easily assimilate open source code, but not all of them are (guess where I come from!). Open source is just that..open...as long as you're comfortable with Unix and can write scripts. 

Neither is a perfect solution, but I'd like to suggest that a quality commercial package with the capability to incorporate (easily) open source code where wanted/needed is pretty close to it, providing a starting point to get analyses rolling while allowing incorporation of in-house or open source codes to modify workflows/graphics as needed for specific users/applications.

Such solutions do exist...but I don't believe this is the proper forum for an advertisement! Thanks for reading...


Richard Carter
Posted: Monday, September 12, 2011 4:01 AM
Joined: 3/8/2011
Posts: 2


I would ask whether the initial question “open source or commercial” is the correct way of viewing the problem. NGS users should not be driven by philosophical approaches rather they should be asking the more pragmatic – what is the problem the user wishes to solve and what options are available to them.

If it is a simple well established analysis and there is a range of open, closed or even “somewhere in between” software options available then other considerations such as resource – do you have a bioinformatics/IT environment available to support one or other solution, compute availability and of course, cost (not just of software but supporting the solution once it is installed) – dominate the decision making.

If the problem is more complicated; then the choices may become more limited, ultimately leading to a roll your own solution in which case the philosophy of open or closed source is pushed back onto the user.

NGS is a rapidly developing field and it will be difficult for players in either the commercial or open camps to keep up. Users will have to continually look at a variety of solutions and must be willing to be agile and versatile in their solutions. An ideal solution allows the user to do the simple things with low overheads right now but leaves a lot of head room to incorporate alternatives as they arise.


HARSHA RAJASIMHA
Posted: Thursday, September 15, 2011 11:01 AM
Joined: 3/4/2011
Posts: 11


Thanks Peter. I think you are right. With Genomatix and similar solutions, the separation between open source and commercial software is shrinking. The ability to seemlessly integrate novel open source algorithms and tools into a commercial package drastically improves the usability of such tools as it comes with the technical support which is often missing or limited in  open source alternatives.
HARSHA RAJASIMHA
Posted: Thursday, September 15, 2011 3:19 PM
Joined: 3/4/2011
Posts: 11


 

Hello Richard,

 

 Thanks for sharing a very practical perspective on this topic.

Indeed, the field requires that the users too stay informed about the latest algorithms and choices. It is unlikely that a single solution will emerge as a winner on all fronts. The ideal software solution for most NGS applications is likely going to be a combination of open-source and/or commercial tools sieved together in a pipeline.

 

 The question intends to seek perspectives on how the current and future NGS data analysis community will look like.

 3-5 years from now, would the NGS user community be mostly using open source software?

 commercial software solution? or

 would a combination of both continue to be the norm as it is today?


Richard Carter
Posted: Friday, September 16, 2011 10:26 AM
Joined: 3/8/2011
Posts: 2


Harsha,

I think it is a brave person who would predict what the NGS community will look like a year from now, let alone in five years time The pace of progress is frenetic and with the new players waiting in the wings unlikely to slow down any time soon. Therefore, I think that it is inevitable that the software will not converge to an "ideal" solution. Rather it will continue to grow organically with commercial software vendors and open source projects appearing and disappearing as their products fit an ever changing market.

A good time to be a programmer and probably an ever better time to be a user of NGS.

Rich


HARSHA RAJASIMHA
Posted: Monday, September 19, 2011 11:01 AM
Joined: 3/4/2011
Posts: 11


Ha ha... indeed, the brave ones are investing in NGS product development with their own imagination of how the future would like... happy

pharmatroll
Posted: Friday, January 27, 2012 10:50 AM
Joined: 4/22/2011
Posts: 1


Is anyone pursuing a "supported open source" model?  The example I'm thinking of is the way Linux is packaged and distributed:  builds that include standard apps,  work on multiple platforms, and are easy to install.  Add some pipeline support and I think you could have a viable business, especially as NGS penetrates to smaller labs with limited support and budgets.
Richard Everson
Posted: Monday, January 30, 2012 1:32 PM
Joined: 3/4/2011
Posts: 1


Both the Broad and Wash U sequencing centers have announced they will make their pipelines available as public domain projects.  Anyone tried implementing them?
HARSHA RAJASIMHA
Posted: Thursday, April 12, 2012 10:35 AM
Joined: 3/4/2011
Posts: 11


Broad's NGS pipelines are part of the GATK software that has been open to the community for about couple years now. Initially, it was not very straight forward and a bioinformatician had a big learning curve to get productive use of it. Its getting better with new versions. Not sure about the Wash U pipelines.
 
Please login to post a new topic or reply to an existing post.
Privacy Policyryrbrbzwubyssrtvufcrtwab|Terms of Use