Forum:What's the general view on point & click bioinformatics for routine analyses?

Question

[Simon]

Full disclosure - I work for a company called Basepair that is a for profit bioinformatics platform provider.

It's founder came from academia where he found he was spending most of his time performing routine analyses and repeatedly answering the same questions related to NGS data analysis & visualization. He set out to create a point & click way of enabling those who were new to NGS or who had less computational experience to analyze, interact and visually explore NGS data on their own with industry standard public domain tools, setting them up to collaborate with a bioinformatician once they had a more informed question. We believe there is a place for this kind of bioinformatics solution, but would be curious as to what others think in this forum?

Anyhow, if anyone would like to try out the platform before providing feedback you can simply sign up for a free six sample trial on our website and let us know what you think. If you want to go on and use it from there there is no upfront license fee - just a pay as you go per sample usage model that includes unlimited analyses per sample and 12 months storage. What's more, if you quote 'Biostars' we'll offer between a 20% and 85% discount depending on the sample type.

[Dunois]

Is there any reason why your organization cannot offer a locally installable solution?

I'm not signing up on your website just to find out what is happening in the backend, but from the screenshots and blog posts (e.g., https://www.basepairtech.com/knowledge-center/a-deep-dive-into-differential-expression/) it appears you've done little more than wrap a GUI (R-based?) around open source tools funded by public (and in some cases, personal?) funding. In this case, what portion of your profits are you using to support these open source tools and to support the development of new tooling in the future?

It is also so tiring and uninspiring to have anything and everything in silico go to the "cloud" with for-profit entities subsequently trying to extract every last unit of profit by monetizing every possible aspect of the users' interactions with the software and data.

[GenoMax]

Is there any reason why your organization cannot offer a locally installable solution?

As stated in the post this is a for profit company. Supporting a locally installable solution (which may need to take into account diverse infrastructure) is likely beyond what the company can do with the resources they have (my speculation). By controlling the backend infrastructure they are able to offer a uniform user experience. Providing tech support is likely significantly more expensive when they have no control on end user infrastructure.

When one has the knowledge and resources to do something it is easy to overlook others who don't have these resources (human or otherwise) available. For them this kind of solution is the next best thing.

[Simon]

Thanks, this is indeed one of the reasons it is challenging to provide something that is reasonably affordable if the offering is predicated on a locally installable solution. This being said however, we do offer the option to plug in an institution's own cloud account into the platform rather than needing to upload any data to ours. I realize it's not the same thing, but it does at least mean that you remain connected to other tools and resources you have there and that data never has to leave your own IT approved environment. Not to mention of course that you can then benefit from any pre-negotiated cloud pricing your institution might have put in place.

Answer 1

[Istvan Albert]

I have story that I sometimes share.

Some years ago I was contacted by a salesman for a company offering RNA-Seq analysis with a point-and-click interface. Back then TopHat was the king of the hill.

We get to chat, and, of course he is heavily promoting the services. I was curious about the price, so as soon there was lull I asked him point blank. So how much does it cost?

He pauses, then says: For an academic institution like yours, it is only two-fifty per sample.

I still vividly remember that strange sense of disconnect from reality that came upon me. I honestly wondered, what is two-fifty? What units are these? Is $2.50 that is so incredibly cheap I would sign up myself. Or is it $250 per sample? That is so ludicrously expensive, one could hire a personal bioinformatician to work for them at those prices.

So I asked him. What does two-fifty mean? Is it $2.50 or $250

The salesman stopped, gave me quite a lengthy, icy silence, as he processed my arrogance: "$250, of course"

The sales pitch stopped shortly thereafter.

PS: a few years later, I checked on the company - it went out of business.

--- end of story ---

There is the answer to your question in my opinion.

There is a need for the services - it is just not obvious how to price it so that it is worth it for both parties. Your website does not quote costs either, apparently, one has to "sign up" to see the pricing ...

I would say academia is not a good customer base for that; the needs are too varied, the data too non-standardized, and the biggest problem is that there are no dedicated resources for third-party analysis.

After all, an academic institution's job and responsibility is to train students how to do an analysis - that is part of the mission.

I would strongly suggest targeting biotechnology industry players.

[Simon]

Great point and story Istvan! I guess you are right, it often depends on what it costs to use such a service. Not sure if you would agree, but our research seems to to indicate that it is generally acceptable to pay an additional 5-15% for bioinformatics all in on top of the cost of generating the data in the first place (sequencing + library prep). But of course I'd welcome other perspectives on that. Interestingly, it just so happens that because we have architected the Basepair platform to be very lightweight & efficient when it comes to deployment on AWS, we find that by pricing our per sample costs in this range (after taking into account the aforementioned discounts) it enables us to put enough back into future development and support of the platform.

[Istvan Albert]

I signed up to check out the site.

I think it is very neat and seems well done, to me it feels like Galaxy done better and in a focused and targeted manner.

I wish you guys all the best.

[Simon]

Also, to answer your comment about a lack of pricing on our website, we actually used to publish our pricing/sample, but found that it was generally having a negative effect on interest levels. Researchers with larger sample volumes that didn't realize we offered volume disounting seemed to be put off. Anyway, to be transparent, these are the prices we charge for each data type on a pay-as-you-go basis for low sample volumes:

• RNASeq & ATACSeq: $40
• Panels: $20
• CUT&TAG, CUT&RUN, SnapChip: $15

[Istvan Albert]

The problem with this type of pricing is that it makes it look like the cost is proportional to the samples. And I fully sympathize with the challenges of pricing in bioinformatics.

You have to sell the product to an audience that has no idea how much something does actually cost ... and where you have to compete with "free but tedious" options.

IMO an RNA-Seq analysis should cost about $1000-$1500 and is fairly independent of the number of samples (within reason - when going from 4 to 24 samples the change in the compute costs are negligible).

The real cost is setting it all up, appropriately accounting for the many unexpected issues that pop up, coming up with the proper experimental design, evaluating the data and results, and making sense of the final results.

And since the process is so unpredictable, one needs a fairly large buffer - the runs that go fast and with no troubles pay for the ones where everything goes bad.

Answer 2

[GenoMax]

You are asking this question in the wrong venue. A large fraction of people here either already use command line tools or are aspiring to learn about how to use them.

That said, there is a place for point and click bioinformatics tools for all bioinformatics analysis. It empowers bench scientists since it can give them a confidence that they are able to do "this". Until your platform reaches a level of maturity that the word gets spread around by reputation of your platform, things will likely remain slow.

Having seen this play out with microarrays (command line-only tools to finally a few commercial tools used by many), expect this to eventually happen with NGS space. One difference compared to is microarrays is that they tapped out in terms of innovation relatively quickly (in a decade or so). In NGS space there is still a lot of new things happening.

If Basepair manages to survive these initial turbulent years then you can probably look at getting acquired by a large existing vendor (and a nice payday?). If you are able to sign up institutions (where you sell them a block of analysis units/time) that the internal users can draw against you may get better traction.

Answer 3

[Jeremy Leipzig]

[Simon]

Thanks Jeremy. If I am not mistaken I think you are saying the same thing as Istvan. That in theory, providing they are used to derive initial insights into the data in order to facilitate collaboration with the analyst, there might be a place for easy to use NGS analysis solutions for those will less or no computational expertise. However, those companies who have tried have invariably failed due to an inability to deal with the variability of NGS data, the time it takes to analyze it and the large data footprint it generates. But let me know if I've got the wrong end of the stick.

[Istvan Albert]

In my opinion, the business model that relies on getting lots of small signups where every user pays a small X amount of money per sample is very difficult to make work in academia.

The only player that IMHO could make that work would be Illumina via Basespace - since the data is already there and people have been paying Illumina already. The barrier of entry is extremely low there.

I would strongly suggest completely refocusing on partnering with institutions/organizations rather than with individuals using a pay-as-you-go model.

I have been providing bioinformatics support in academia for over a decade; I have never heard of a bioinformatics support center at any university that would completely self-support itself with all costs born by the "customers".

[lethalfang]

I agree with Jeremy. The biggest bioinformatics problem for bench scientists is not that bench scientists don't know how to use CLI. The obstacle to CLI is easily solved with GUI wrappers. Many companies think if they solve the CLI/GUI problem, they'll solve the bioinformatics problem for bench scientists. But that's not the problem. It's the interpretation of the (massive) data that requires bioinformatics research expertise.

Answer 4

[brian.fristensky]

The BIRCH system is a free collection of common bioinformatics tool unified through a programmable object-oriented GUI known as BioLegato. BIRCH currently includes 14 BioLegato applications, each specialized for a different type of data. For example, blreads is a BioLegato application for tasks that take NGS sequencing reads as input. This screenshot shows three sets of paired-end Illumina files (trimmed by Trimmomatic) to be used as input for error correction.

The corrected read files would appear in a new blreads instance,

Because output appears in a new BioLegato instance, there are no dead ends. You click your way from one step to the next, and each output window becomes a new jumping off point for whatever you wish to do next.

BIoLegato applications are included for DNA (bldna), protein (blprotein), DNA alignments (blnalign), protein alignments (blpalign), Entrez queries (blncbi), among others.

For a better idea of how BioLegato works, see the BioLegato tutorials, or visit the BIRCH YouTube channel.