HiSeq2000 - Next Level Hacking


#1

We got a HiSeq 2000, Next Level Sequencing Machine from the Genomics Facility of Department of Biosystems Science and Engineering in Basel. Contact through Biozentrum, University of Basel. It’s a powerful high-throughput sequencing system that enables large-scale genomics for a broad range of applications. Now this type of machine seems to be quite difficult to get up and running and also reagents, flowcell-kits and software licences can be expensive. Since more of these machines seem to show up in second hand (there are new machine generations by Illumina) it would be worth trying to find a way to make them work. Sequencing for all.

Let’s discuss ideas and plan a hack session for working on the machine.

For pictures see the wiki: https://www.hackteria.org/wiki/HiSeq2000_-_Next_Level_Hacking


#2

Hi,
I saw your write-up about HiSeq 2000. I also tried to do some reverse engineering on this sequencer at Hackuarium (the machine is disassembeled now for parts). I mostly worked on understanding the electronics and the control software. Did you have any progress since? Is there anything I can help?

Masoud


#3

Hi Masoud

Good to hear from you. I saw pictures of your disassembling. So far I tried to understand how that machine works and how everything connects. I just now got the WorkStation (Dell T7500) that goes with it and try to make the software run (all hard-drives with software were removed). Did you get the computer and software with your machine? Then I was also looking into making the individual modules run to eventually run an open software on it. I think the pumps and valves should not be a problem. What I am most worried about is the FPGA-bord developed by Illumina to control the XYZ table. Did you try to communicate with that? Any information would be helpful. I hope to be able to reverse engineer it with the machine running on original software. Also a bit worrying is the camera frame grabber. It’s a commercial card (by Active Silicon) but the Developer Kit is not free I guess and then they also seem to use line cameras with Time delay and integration (TDI) on the machine…

What do you think? Would love to come visit you and look at the components.
Urs


#4

No, ours didn’t come with the workstation but I have the frame-grabber.

Yes one challenge is to make the frame-grabber working without the official drivers.
Hamamatsu provides dcam-api (https://dcam-api.com/) but I doubt it also provides drivers (I am going to test it as soon as I find a PC with PCIE slot). There is also Aravis (https://github.com/AravisProject/aravis) which enables interfacing grabbers with GeniCam interface. In worst case scenario one can use another CoaxPress grabber.

By investigating the files in the control software it seems to me that the workflow and pipelines is sent through the USB-Serial interface in text format and then the embedded system takes care of fine-grain control. This and next week I am going to confirm whether we need to understand XYZ control interface or not for being able to use the machine.

You are welcome to visit us in Renens.

Masoud


#5

Nice. There is even a TDI example in the Hamamatsu API. Looking forward to your further findings.
Urs


#6

Hi, My company Survival Research Labs, recently received the components from 120 HiSeq2000 units from Illumina. We are selling all components for cheap on US ebay. Cameras, optics, etc. See SRLSALES for components. Please contact mark pauline @ markp@srl.org for bulk sales.


#7

I got one unit of the CCD controller and two CCD heads (C10000-512). Does anyone know how to connect the powers? I saw there are three inputs. One is CameraLink, one BNC, and one DB9. I guess DB9 for the communication. How is the power supplied?


#8

The yellow question mark in the middle is actually the CCD head.


#9

Hey!

Would be interesting to have some proper documentation photos of the different PCBs and other hardware for more people to be able to help out understanding the beast. So for PCBs that would be pictures with good lighting (from several directions), taken from straight above and in focus over the whole board so that markings on chip are readable, and of both sides of each board.

What I can see on this is the following:

Bottom left there is a Renesas HD64F2367VTE33V H8S 33 MHz 8-bit microcontroller with 384 kb flash and 24 kb ram.

Above that there is a Microchip 24AA64ISN, 8kb EEPROM

Further up there is on the left an Altera EPCS16N which is a 16 Megabit (so 2Mbyte) configuration memory for the FPGA that is the big one on the right. So this would contain the encrypted bitstream that the FPGA runs, the FPGA itself has no persistent memory.

The Altera CPM3032ATC44-10N is a small CPLD programmable logic probably just glueing things together.

The big chip is an FPGA Altera Cyclone II EP2C20F484C7N - 65 USD a pop on digikey - 1172 logic blocks, 239k bits RAM, 18752 logic cells. This is big enough to run a CPU that can run Linux but more likely it is acting as DSP for image processing if it is a camera module. Also you see 5 pairs of traces go at constant distance down to the connector bottom left, that would be some differential high speed communication (like ethernet, usb, sata, hdmi etc)

To the right of the FPGA there is a 32-bit wide SDR SDRAM memory which is 64 Megabit so 8 Mb and 6 ns access time or 167 MHz.

Also you can see that the whole board is sprayed with testpoints, looks like basically every pin on the H8S microcontroller e.g. has it’s on testpoint in a via.

Cheers,
Bengt


#10

!


#12


#13

Nice pictures Daniel and analysis of the PCB Bengt.
Here some more pictures I found on ebay (have not disassembled our unit - yet :slight_smile: )

Hamamatsu S10405 Line CCD Cameras with water cooling system:

HiSeq 2000 Genome Sequencer Controller:

Xilinx Spartan Image Control Circuit Board 5014697 Illumina HiSeq2000:

Illumina Driver Circuit Board Assy. # 23061 f/ Illumina HiSeq-2000:

I reversed all the ports and many commands on the hackteria wiki here:
https://www.hackteria.org/wiki/HiSeq2000_-_Next_Level_Hacking#Comands

There are commands to control the x-y-z motors plus 3 tilting motors and to do the TDI (time delayed integration) scann… so we should soon be ready to do some high res fluorescent imaging. The thing can scann a full microscopy slide at a resolution of 16GIGA pixels.

Urs


#14

Its basically a microscope? Fluorescently labelled organelles?


#15

Sequencing by Synthesis - basically chemistry/enzymes that fluoresces in different colors depending on which base is added and millions of strands being synthesized in parallell in one microscopy frame.


#16

It’s a fluorescent microscope with picture scanning. This is one segment of a slide (flow chip) that we scanned last weekend:

You can see the DNA clusters (dots) and some other crap (did not pay much attention on cleaning). Don’t know what the different lines are…

And it is just one tile of the whole image. The slide is scanned with 6x16 tiles per lane and 8 lanes.

The pictures of each tile are 2000x10’000 pixels at 16bit :slight_smile:
Let’s put some tardigrades in the flow channels and make a huge poster…
Urs


#17

The DB9 in the camera board is for power. In the HiSeq controller (as shown in the image above) has the DB9 power output for the camera. If someone with the controller can measure the voltage output of the DB9 pins, one can find out the power required to drive this camera control board.


#18

Christopher Burge at Stanford University in California hacked an Illumina sequencer to measure protein–DNA interactions at very high throughput. Nice.

“The GAII is basically a pumping system that pumps things onto the flow cell and then onto a fancy imaging system,”

https://www.nature.com/articles/d41586-018-05769-8


#19

This project is building a board for sequencing using Ion Personal Genome Machine chips.


#20

First of all, thanks for documenting what you have done. It’s a fantastic resource. I’m interested in re-purposing the microscope from a HiSeq2500 and was wondering if you have made available the control software mentioned here:

With many of the commands known it is possible to use the HiSeq2000 as a transmission or fluorescent scanning microscope. We developed a simple control software based on PureData to send commands to the HiSeq control hardware and get images from the scanning microscope. To view, set-up and read the scanner cameras we used HCImage software and TDIScan tool both available for free from Hamamatsu.

I have found a couple of engineering students to work on it over the (Australian) summer holidays and if they could use what you have already done as a starting point that would be awesome. Details of your illumination for transmission imaging would also be good if you have anything you can put up. My goal was to see if they could get it integrated into https://micro-manager.org/ which would make it accessible to the microscope community more generally. Any thoughts you might have on what problems we may encounter would be much appreciated.

John


#21

Hi John

Great to hear you want to re-purposing your HiSeq2500 together with your students. I updated the wiki with a link to the PureData Sketch and the scan sequence. It’s a hack and not easy to get to work. Also you will have to see if it’s compatible with the 2500. Feel free to ask if you are stuck. A better, easy to use software would be great and I am happy to help. All the commands we need are here I guess. You will have to call DCAM API commands to do the picture grabbing (not sure if this is possible in Micro-Manager).

Urs