HiSeq2000 - Next Level Hacking


#42

It looks like everyone is available Friday 6/11 1 am Melbourne time. Shall we lock that in?


#43

friday 16th? sounds like a plan…


#44

works for me, in my calendar.


#45

Looks like the only option for everyone to meet is Thursday at 14:00 GMT actually. I have locked that in and it should have sent you a calendar invite. Let’s use this Jitsi room at that time: https://meet.jit.si/hiseq

By the way, I just received an email from WeMakeIt and the Science Booster match funding is still available early 2019.


#46

We had the call today joined by Oliver Keller, Bengt , Urs, Greg and me.

  • We asked Bengt a few questions and we studied some pictures from the wiki. Greg concluded that while it has a lot of potential and is really interesting it’s probably not suitable for Our Sci work.
  • We were trying to figure out how many HiSeq2000/2500 there are. Bengt mentioned omicsmap.com (down right now – but looks like below)
  • Urs talked about how the pure data sketch worked and we re-iterated everything we said about the camera and the frame grabber. Still not sure how to integrate this part of it into the software plan.

Actions:

  1. I am buying a frame grabber, @gaudi will send me a camera. This should allow me to see how hard it is to integrate with the rest of it.
  2. @gaudi will announce a get together in early January, somewhere with a machine, probably Switzerland

#47

On the call today we talked a bit about the number of HiSeq 2000 machines manufactured so I tried to find some information about sales and such.

According to this document there are ca 2300 units shipped from the HiSeq platforms in total - that includes HiSeq 2000, HiSeq 2500, HiSeq 4000 and X. (there was HiSeq 1000 which is same as 2000 but with 1 flow cell instead of 2, same with 1500/2500 and 3000/4000 but I assume they are all included in the numbers). Howver 1000/2000 could be converted to 1500/2500 at least after a serial number 1400 or so - I did also see a number of figures of units installed of Hiseq 2000 from 750 to just above 1000 but all from 2010/2011 on broken and forgotten websites…

The HiSeq 1000/2000 was announced January 2010, HiSeq 1500/2500 in January 2012(with v4 chemistry for it launched in 2014), HiSeq X in beginning of 2014 and HiSeq 3000/4000 beginning 2015. If I understand correctly the HiSeq 1000/2000 was announced to be discontinued in fall 2013 but I didn’t find when they actually stopped selling it.

According to AllSeq and Illumina investor presentation from May the cost of HiSeq 2500, the successor of the HiSeq 2000, was 740,000 USD, the HiSeq 4000 is 900,000 USD and the X Five is 6M USD for a pack of 5(!) and the X Ten is 10M USD for a pack of 10(!).

http://allseq.com/knowledge-bank/sequencing-platforms/illumina/

You can also see in the Applications Table for the HiSeq 2500 at allseq that there is a whole range of tasks it can do (even if not optimal for some of them). Couldn’t find any such information specifically for 1000/2000 yet though but at least it is the same tech and structure, if I understand correctly the Hiseq 2500 mostly being a bit faster and being able to do 2x125bp reads in certain cases instead the max 2x100bp reads on the older ones.

When the Hiseq 1000/2000 was launched it claimed to be the first platform getting sequencing cost of a human genome below 10,000 USD - a full single run costs 23,000 USD in consumables though… WIth the HiSeq X Ten you supposedly get to a consumables cost of “significantly less than 1,000 USD per genome”.


#48

Sorry for my bad connection on the call.

Thank you @kaspar for summarizing.
Thank you @bengtsjolen for the sales numbers. I think >1000 of these machines is worth the effort. And the unit I have is absolutely available for the community to work with and I am sure we can manage to get at least one more with a successful campaign. And @jmarkham and @tboysen have their machines too.

I uploaded a zip with a log files and extracted command from my working folder for you to get an idea what this looks like here:
http://www.gaudi.ch/GaudiLabs/wp-content/uploads/HiSeqLogs.zip

The log file is from a real sequencing run that we made. You can see the communication with the machine, that contains the commands and also the sequence in which the commands were sent to the different units plus lots of extra information. :slight_smile:

Will send you the camera @kaspar . However as you do not have the FPGA unit yet you will have to somehow power the cameras. This is by a DB9 connector and we need to figure out the power pins (has been asked by @forest earlier in this thread ). I will try to figure this out.

Then I made an estimation of what I think is need to write this software (I do not want to do it though :slight_smile: ). I estimate its 1 month of work, 160 hours. To get to a working software including the camera interface. (1 day research, 3 days of testing with camera sdk, 1 days of testing with serial interfaces, 2 days setting up environment, 2 day interface design, 6 days implementation, 3 days testing refine, 2 days extra) something like that.

Then I think we should be able to raise like 6000-7000 euro including the science booster (2k from GaudiLabs + 1-1.5k from others). Putting aside 500 euro for materials etc you can calculate the hourly rate. This does not include cost (travel) for our gathering.

Then I was asking myself what could be offered as “rewards” in an open software camping. I think we should work on the code openly from the beginning and not offer “early access to code”. I checked some other successful projects and here is what I found (in increasing order):

  • A big thank you.
  • Sticker (maybe ChinaGOSH)
  • A link to your website with your name in the supporters section
  • Top level thanks in the project credits, announcement emails, website, face book and twitter, what ever.
  • Big Logo on website and in the About / Splash page
  • Paper Project Sponsor Certificate (think this can be attractive for some institutions)
  • Eligible to vote on features / design decisions
  • Propose Features to be implemented
  • Take part in monthly development hangouts
  • Influence over direction and architectural decisions

hope this makes sense.
Urs


#49

Sorry, because i was really overloaded with my work here i missed the whole thing with the phone conference and all. And at the moment i’m on travel so i can’t react quickly – e.g. doing the measurements and documentation to power the cameras.

Aaah, if i had forseen this project i had left all machines complete! But the space problem in the institute was pressing back then. And to haul them in one piece at home on the attic was not really an option… But one is left and most of the components. And more HiSeqs will come in the future - the moment the next machines are decommissioned here i will get them complete - including the PCs.

Writing the software to handle all the UART links i think is not a big hurdle - done this before with python to control lab machines build from parts of the ABI and Illumina components. Reuse of the cameras was sometimes nasty because not in every case drivers and SDK are avaliable or it’s costly so that reusing them involves the manual integration of libs, binaries and registry hacks. I’m eager to see how the TDI functionality can be implemented with the Hamamatsu SDK.

When i’m back in office next week i will search the Phoenix cards and stuff. Maybe i’s helping if i send the whole set of components needed to test the imaging hardware/software to someone doing that part? Why treasure up these things if i have not the time at the moment to experiment with it.

Best,

Teide


#50

Thanks for the estimates @gaudi. Integrating the camera/framegrabber is still the biggest known unknown. There are likely other unknowns. The estimated hourly seems fine for me though, depending on how much we spend on travel. We can review and revise as we progress.

That would be good. I was about to order the frame grabber. Let me know if you have one you can send. You can discuss with @gaudi who should send the Hamamatsu camera.

By the way, has anyone used LabView with the HiSeq2000/2500? What does that allow you to do? Does that integrate the camera?


#51

Hi all, first I must apologise for missing the meeting, which was just due to my hopelessness. I had 1 am Friday in my mind which then converted it to Friday night (and hence a night too late).

I’ve tried the micro-manager mailing list again in the hope that they will release the driver source to someone wanting to do development.

@gaudi, I fly out of Barcelona on 12/1 and potentially could meet up before then. If you send dates I’ll try to organise around it.

@kaspar, Hamamatsu say they support LabView so I think you’d need to install dcam-api driver on Windows and fire up LabView and see what it could find. I expect that other LabView users in engineering/physics areas would use line scanning so you might get lucky.

Has anyone tried using micro-manager with the HamamatsuHam device manager just to see what it can find?


#52

Just to clarify, what’s the final output that is sought? Is it a piece of stand-alone software to drive the instrument or is it device adapters etc for micro-pilot? If the former, is there functionality required that’s not in micro-pilot, which has already implemented the microscopy features that most people would need. It also allows device adapters to be written for syringe pumps etc and support for synchronisation. In principle it could be used drive all the components.

Also, was the plan to use the dcam-api? When I downloaded it I had to click on an agreement. I didn’t read it but I suspect that this, or similar, is why micro-manager can’t make the source code publicly available for the HamamatsuHam device adapter. Should it be an issue there may be ways around it, but before making enquiries I thought I’d check.


#53

Hoi Zäme

I measured the HiSeq camera connector and this is what I found.

I can send you the camera and a connector @kaspar , just send me your address. @tboysen can you send kaspar a frame grabber card and maybe a CameraLink cable?

Then for that get together I suggest a HiSeq session in January @ GaudiLabs, Lucerne Switzerland. (CCC is hard to get tickets for all and bit far for me to get the HiSeq there I guess)

I think about three days would be good, so that we can:

  • Make work the hacked machine
  • Experiment with fluorescent scanning and other potential applications
  • Think about software, interface and implementations
  • Design crowdfunding and shoot a video

I can probably find cheap places for you to stay here (in the lab and in my or friends place). Not sure how we can do about the travel cost yet though.

There is also possibility to go to Zurich, @dusjagr forest house or individual sessions (skiing) before or after.

What do you think?
Here a doodle with two dates:


#54

I tested the connector to the camera and there was a mistake in the first diagram that I uploaded (now corrected). There are (of course) not two 15VDC but a + and a - 15 VDC. When I connected the 15 VDC to the -15VDC a fuse in the camera blew up (yes, these are well designed devices with fuses that hold). And that’s how I realized and measured again and found the mistake and tested with +/- 15VDC and now it works.


#55

The card is ready, the Phoenix D48CL – still in search for the
corresponding cables. Is only the cable needed
Grabbercard<–>Instrument? And of course i have to know where to send
the package.

Teide.


#56

@tboysen , @bengtsjolen , @danield (and everyone on this thread) you are happily invited to join the HiSeq Session in GaudiLabs. Please fill in the doodle if you feel like joining:

Urs


#57

@kaspar the camera is on its way to Bristol. Hope you get it well.
Urs


#58

Ah, I just ordered for a ChaosLooper. Was hoping to get that order in before you sent it so you could include it. :grimacing:


#59

Hello All,

Just wanted to introduce myself, my name is Will and I am a research engineer in the innovation lab at a sequencing center in NYC. I have been perusing this forum for a short time now and am very interested in your efforts in re-purposing some HiSeq instruments. We will have a significant number of HiSeq 2500’s decommissioned over the course of the next year or so and have been given permission to do what we wish with them. We plan to put them to use for high-throughput experiments as typified by recent work of the Greenleaf lab at Stanford. We primarily have capabilities in microfluidics (small cleanroom for fabrication on site) and single cell technologies/expertise. If there is anyway we can contribute to the efforts please let me know.

Thanks,

Will


#60

Package arrived yesterday, camera module looks very scientific.


#61

Card and cable is also on the way.

Best,

Teide