HiSeq2000 - Next Level Hacking

The HiSeq session in GaudiLabs Luzern is coming to an end. @kaspar @jmarkham @bengtsjolen and I worked on the development of a new open source software for several days. I would call the session a total success. We had great synergies and got a lot of important things done. We looked at the existing hack and PureData software. We compiled the DCAM-API sample codes in Visual Studio and got images from the camera. We then even got to compile a version of the Micro-Manager DCAM device driver. We discussed the implement of the HiSeq camera control as a Micro-Manager device adapter and add plug-ins for lasers, optics etc. Finally we made video interviews for the crowd-funding campaign that we plan to launch soon.


Link to the new ReSeq project page.

3 Likes

Thanks @gaudi for your warm and generous hosting and your enthusiastic support for the project. And thanks to everyone for their efforts. It was a lot of fun and I’m really looking forward to getting things happening.

J.

Thanks @gaudi and family for hosting us!

In the last day we worked on the video and remote access (we now have FTP and SSH access in addition to the VNC thing @bengtsjolen set up).

We just hit Hacker News front page as well and are getting some more sign-ups to the email list through that. People are also posting some interesting links to similar projects.

https://news.ycombinator.com/item?id=18896318

1 Like

In 2006 and 2007 I worked with a group that was attempting to build and open source DNA sequencer. I wrote much of the system integration code, linking the camera, fluorescent illumination systems, stage motion, and microfluidics. I always thought we would have had more impact if we focused on the general purpose applications of the system. It’d be very cool to help out with this project. I hope to get my hands on a similar system sometime in the coming years. I did work with a Hamamatsu camera, but it might have been of a different type. In any case, we were able to get everything going on Linux. Maybe I could be of better help if I found the old source code.

Here the first cut of our campaign video.
Two minutes is really short, we had to cut out a lot of nice footage. What do you think? Corrections?
The picture of the sequencing machines is from the internet. If you @tboysen have a short sequence I would be happy to put that instead.
And @jmarkham can you send me the picture (or a short video) of your microfluidic system so I can put that in as well.

4 Likes

Thanks Urs. Looks good. I’ll try and get something tomorrow. In youtube the resolution/framerate is 854x480@25FPS. Is that what you want or are you editing it at something higher than that?

I think I’ve posted these articles, but if you wanted, you could do some Ken Burns rendering of the figures. Although possible we already have enough visuals.
Nature article about re-use of GAIIx machines:
https://www.nature.com/articles/d41586-018-05769-8
Application to measuring on/off rates for CRISPR/dCas9 do DNA clusters on the flow cell:

Here is the nanostring product that I mentioned. They immobilise RNA on a slide and use a microscope to count molecules. By requiring the binding of two things - the capture and reporter probes I guess they reduce non-specific binding. Presumably they use some sort of amplification reaction in order to see single molecules, however I speculate that some of the new fluorophore constructs may offer an alternative.
https://www.nanostring.com/scientific-content/technology-overview/ncounter-technology

They can also measure DNA so if one could stick down the captured molecules well enough, then a kind of DIY sequencing-by-ligation might be possible by flowing through cunningly-designed (ie organism and assay-specific) sets of reporter probes and carefully measuring where they bind to each stuck-down molecule.

HD 1080i 25fps would be ideal. 854x480@25FPS should work too. Guess we should add the links to the articles to the campaign text.
Urs

I was gonna write an email for academic/institute mailing list distribution, including an FAQ, which I’ll send around for comment. Possibly the latter could also be posted on the wemakeit.com and or reseq.hackteria.org. Do we need some other copy written for either of those two sites? I’m happy to do so. I was gonna look at some wemakeit.com projects for prior art to inform style and content. @kaspar, did you set up a wemakeit.com project?

Qee and James have done some documentation on cabling and layout in the HiSeq 2500. It would be good to make that plus some documentation on how to split and re-house the innards available to others. @gaudi, are you happy to have that in the wiki? If so, do we need editing access?

Ta,

J.

Thinking ahead, an open-source sequencer would make it possible to develop algorithms to address some of the current machine’s limitations:

Currently, this tends to be done downstream of the original image processsing. For example:

I wonder if ML-based image segmentation and classification methods applied to the image stacks directly might do better. In this context stack would be the series of images from a sequencing run in the neighbourhood of a cluster and classification would be imputing the DNA sequence of the original molecule from which the cluster was derived.

Video is looking good @gaudi

  1. Can you put my title as “Independant Software Developer” or “Electronic Engineer and Software Developer” or just “Software Developer”?
  2. Do we have a better take for the final bit? It sounds so terribly scripted compared to the rest :laughing:
  3. Probably obvious, but the cuts could still be smoothed out a bit.

@jmarkham More copy would be good. Could you write something about potential applications? I started the project page and requested you be added, go to https://wemakeit.com/projects/reseq-making-the-hiseq-open/edit and see if it works. You could also throw something up on an Etherpad or Google Docs in the meantime.

We now have 31 people on the mailing list and hopefully should have some press in a popular blog soon. I also found the source for omicsmaps.com (still down) and opened some issues on there. https://github.com/nickloman/omicsmaps/

I was also wondering if we could wrap up the bottom illumination into a nice little circuit board and offer it as a reward and write a Micro-Manager adapter for it?

I’m back in the office…

I can provide a sequence of the machines here - some HiSeqs are still
operated, not soo many as seen in the short shot. I will do it today.

Best,

Teide

@kaspar the wemakeit link won’t work for me. Initially I get:

Private function

Are you authorized to access this tool? To get access, please register here.

Even though I am signed in. Then when I sign in it just goes to the default opening screen and no sign of hiseq, Also it doesn’t come up when I search there. My profile name is jmarkham and my login is my gmail address. Which did you share it with? I’ll do some copy.

Did Nick Loman contact you? What’s in his code?

I like the idea of an illumination board. I think there is an arduino board with a MM device adaptor that is used to retrofit triggering of some sort to older scopes. It may be worth looking at that. Also there are spare serial ports on the main box available for use (on mine anyway). Ideally, you wanna use a monochrome source at one of the fluorophore emission frequencies so that you can use it with the remaining three fluorescent channels. Even better would be to be able to switch frequencies and to adjust the intensity. Independent illumination from above may also be useful - at least I have found it to be so in some applications. That’s possibly another discussion though.

@tboysen the package arrived! Thanks for that. I haven’t had a chance to do any unboxing it yet.

The code is a Python web application including database dumps from 4 years ago. If you remember from before, Omicsmaps is a site that maps out the location of all next generation DNA sequencers including HiSeqs. I am hoping Nick will respond soon so we can get some more up-to-date data and can help keep the site up and running.

Regarding the illumination board, to sum up, ideally it would be:

  • adjustable RGB LEDs from above and below
  • easily mountable inside the machine (@gaudi and others who have a machine to hand, any ideas?)
  • easily connectable to the main machine (USB?)

Emission is (558,610,687,740) which is roughly (green,orange,red,ir):

RGB would get you two channels and blue may get through to the 558 channel depending on the optics.

Transmission illumination can be very low level so a cheap little LCD display underneath may work and would give some extra options (as in angle of incidence and polarisation measurement - illumination would be polarised so a polarisation filter on the detector optics may show up something - I speculate).

I will look at the geometry when I go in but from above I suspect you can clip something small onto the objective. I’m not sure if the LCD idea would work so well there because you need a clear path from the objective to the sample. But you never know…

A more conventional method would be to change the optics but I don’t see that being nearly as much fun.

If you make it RS-232 then you can run the wire to the control box rather than to the PC. I guess that would reduce its usability elsewhere though and you probably get USB for free on whatever you use.

J.

@gaudi: Can you define what footage you need? If you just want to replace the two seconds of the muli-hiseq photo i can send you one like this one (with better light?) or a short video of a passby - this i can put you on one of our sftp servers to pick up.

@jmarkham: Thanks for the info! Hope the valves are extending the capabilities of the FrankenSeq :wink:

@kaspar: Hope Grabbercard and cable arrived meanwhile?

Best,

Teide

Hi Teide

Thank you for your support of the project. Yes, it’s about the short muli-hiseq sequence. Should just show what a sequencing facility looks like. A scene with old and new machines like the one you sent should be perfect.

Urs

Yes it arrived just before Christmas, thanks!

I made a short sequence, matching the few seconds appearance in the video, something like that?

Missed it but the Hackaday article went up a few days ago!

1 Like