I've been doing bioinformatics and genomics for almost 10 years mainly using academic software, mostly with command-line interface, as well as writing my own scripts using Perl/python/R.
Lately, I've run into three commercial tools related to bioinformatics and genomics, namely: Golden Helix, DNASTAR, and geneious.
Although I haven't really tried them, they appear to be somewhat similar to one another.
If anyone here has experience with any of these tools (or others that I didn't mention), it'd be interesting to hear what you think. Are they any good? Are you still using them? How well did they integrate with your work etc.
Have these tools ever been compared to one another and/or to academic software?
In general, this thread is supposed to be an open discussion about relevant commercial tools, so feel free to chime in with any thoughts or comments on the topic.
I have tried GoldenHelix more than 10 years ago, because they are a local company and have been offering demos. They seemed fine at the time, and I suspect they are only better now.
My experience with other commercial software is that they generally cost too much to renew, and usually have some kind of license limitation as to the number of computers that can use it at the same time. They update and fix things on their own schedule, which is to say that they aim to satisfy the problems of majority rather than individual customers. If one's problem is exotic enough, it may never be fixed, or take too long. Open source programs can always be fixed directly assuming at least a moderate background in coding.
I don't know why you are looking into commercial software, but the only reason for me to switch from open source programs to commercial would be if I was forced to do so by my university or company. Even if I was creating a commercial product, I'd much rather pay to open source developers to license their solutions.
Thanks for the interesting answer. I am also using open source tools only, but I feel that this type of work comes with a cost regarding streamlining of processes. For example, one may need to use 5+ different tools to complete a standard task such as variant calling from short reads (e.g. BWA --> samtools --> bcftools --> GATK --> IGV...). My personal experience is that much of my time is spent converting and fixing formats so that the output of one tool matches the input of another. I really think the community could benefit from a general platform that will manage files and include many of the common genomics tasks. But I definitely see your point.
There are already existing solutions for pipelining tasks and workflows, e.g. Nextflow (and the excellent, free, community built/maintained pipelines in nf-core), snakemake, Galaxy, CWL, etc.
Most tasks are not unique wheels and have been built before.
I expect the percentage of computeral bioinfologists that regularly use commercial research software is likely approaching single digits. SaaS models are not something most enjoy and put you at the whims of the code managers.
The installation infrastructure is woefully inadequate. Currently, we can't just download a program and run it. Even less so when, as you mention, we need to run several different quite complex pieces of software.
As containerization progresses, we may get to the point where the containers are practically invisible, and a program downloaded as a single executable binary will just run after a download with, say a single click, no matter what operating system or hardware one has.
When that time comes ... and that may be five years out, maybe ten - bioinformatics can radically change, and we can focus on interfaces rather than puttering around with command line setup and installation.
OK, so what you're saying is basically that the high price is not worth it because you can do the same or even better with ggplot (or whatever)? In that case, it sounds like the software is simply not good enough. If they could save you hours of work with one click, then maybe you'd be willing to pay. Also, if they included "hooks" or some kind of API for people with coding skills then maybe they'd be more useful. WDYT?
IMHO, all commercial software are meant for biologists who do NOT code. And if those who code are using commercial software, then it is going to be farce in the near future. After all, many commercial software rely on OS codes.
If the commercial software was no good and/or if it was not producing results that made sense then over time even biologists (who are more likely to use these tools) would stop using them. Most of these software packages have been around for a decade or more so that perhaps says otherwise.
I am not going to get into the debate of commercial/open-source software since each as its place. I have plenty of experience supporting both and they can co-exist in an institution. Choosing software for personal use as opposed to supporting tens/hundreds of users are two very different things and require different thought processes/decisions. If you are asking this for your own use then using open source is a no-brainer, especially since you are a bioinformatician.
You can do your own comparison of the results a commercial tool produces with the open-source alternatives you are familiar with and make an informed decision. When you go down the path of commercial software you are getting locked into the company's development schedule and are going to need to assume some opacity about how some things work under the hood. Most companies would be open about the algorithm they are using, if not their precise implementation. Some commercial software has co-opted open-source e.g geneious uses many of the BBTools, Partek Flow has STAR etc. This may be a win-win for both parties since the open-source developer likely nets some royalty money.
There is a net cost for doing business. It applies to all the work that happens and tools you use. You (or your supervisor) needs to decide what makes most sense for your institution/enterprise.
Besides the list of software you had mentioned there is also CLC Genomics workbench/Science server from Qiagen and Partek Flow to add to your list. Especially if your needs extend to large (hundreds of samples) datasets. Since you mentioned variant calling, cloud based platforms like DNANexus, Seven Bridges may also be contenders.
BaseSpace can also be used independent of the sequencers but institutions likely have security requirements in place that may limit use of cloud services like BaseSpace, especially with human data.
You can also check out SciDAP (https://scidap.com). This platform combines open-source containerized CWL-based reproducible pipelines for processing data with a user-friendly web interface for exploratory analysis.
Thanks for the interesting answer. I am also using open source tools only, but I feel that this type of work comes with a cost regarding streamlining of processes. For example, one may need to use 5+ different tools to complete a standard task such as variant calling from short reads (e.g. BWA --> samtools --> bcftools --> GATK --> IGV...). My personal experience is that much of my time is spent converting and fixing formats so that the output of one tool matches the input of another. I really think the community could benefit from a general platform that will manage files and include many of the common genomics tasks. But I definitely see your point.
There are already existing solutions for pipelining tasks and workflows, e.g. Nextflow (and the excellent, free, community built/maintained pipelines in nf-core), snakemake, Galaxy, CWL, etc.
Most tasks are not unique wheels and have been built before.
I expect the percentage of computeral bioinfologists that regularly use commercial research software is likely approaching single digits. SaaS models are not something most enjoy and put you at the whims of the code managers.
The installation infrastructure is woefully inadequate. Currently, we can't just download a program and run it. Even less so when, as you mention, we need to run several different quite complex pieces of software.
As containerization progresses, we may get to the point where the containers are practically invisible, and a program downloaded as a single executable binary will just run after a download with, say a single click, no matter what operating system or hardware one has.
When that time comes ... and that may be five years out, maybe ten - bioinformatics can radically change, and we can focus on interfaces rather than puttering around with command line setup and installation.