Miseq Network Card

[itguy9]

Hello I would like to know all the different models of Miseq instruments and if any of these models are capable of having a 10 gigabyte NIC card for faster data transfer to one of our network shares.

[GenoMax]

I don't think you want to get inside a MiSeq (unless you are an Illumina engineer). There are no accessible slots to put a 10G NIC either.

We regularly collect data on a network share from our MiSeq's during run and the gigabit ethernet port in MiSeq is plenty fast to allow saving data in real time.

[itguy9]

Miseq Hardware Specs

GenoMax, thank you for your response. So there is no way of replacing the 1gig NIC card at the back of the Miseq with a 10gig NIC card ? Also I notice when I go to Control Panel, network and sharing center I see two NICS in the Miseq, is there an internal NIC card that is used for servicing the instrument ?

[kcchan]

The computer is buried inside the instrument and is definitely not user accessible. Opening it up is probably going to void your warranty.

You can change the config settings to output the data directly to your network share instead of the internal drive. They only generate about 10-15Gb of raw files per day so a gigabit connection should be plenty.

[WhatsOEver]

They only generate about 10-15Gb of raw files per day so a gigabit connection should be plenty.

Just out of curiosity: How do you get that amount of data per day? Our MiSeq requires 2.5 days to complete a 2x300bp run and shorter libraries do not produce that amount of data?!

[kcchan]

Just out of curiosity: How do you get that amount of data per day? Our MiSeq requires 2.5 days to complete a 2x300bp run and shorter libraries do not produce that amount of data?!

Sorry I was referring to 10-15 Gigabytes of data, which includes all the FASTQs, bcls, and thumbnails.

[WhatsOEver]

Sorry I was referring to 10-15 Gigabytes of data, which includes all the FASTQs, bcls, and thumbnails.

But this is not the amount of data that normally needs to be send over the network, is it? I mean, you will only need all the other stuff (bcl, pics, etc.) if your data looks somehow weird. If the fastqc's pass our read QC, we actually delete the run folders.

[GenoMax]

But this is not the amount of data that normally needs to be send over the network, is it? I mean, you will only need all the other stuff (bcl, pics, etc.) if your data looks somehow weird. If the fastqc's pass our read QC, we actually delete the run folders.

Some of us do not use the on-board analysis software for MiSeq, so we transfer (in fact save in real time) the entire raw data folder to a network share for off-line analysis with bcl2fastq. With some longer/well loaded runs this data can be close to 30+G.

[WhatsOEver]

Some of us do not use the on-board analysis software for MiSeq, so we transfer (in fact save in real time) the entire raw data folder to a network share for off-line analysis with bcl2fastq. With some longer/well loaded runs this data can be close to 30+G.

And what is the advantage of doing so? I thought "GenerateFastq" is the smallest, possible workflow I can use for a run? In that workflow fastqs are generated by applying bcl2fastq internally after the run has been completed. At least that is what the technician told us when the machine was installed (at least my ~2 year old notes tell me so ).
I also compared once self-made bcl2fastq output to the existing fastqs in the output folder and there was no difference. Currently, I would only perform a conversion myself if the used indices are diverse and a lot of reads were classified as "undetermined" to "rescue" some of them (but this never happened so far).

[GenoMax]

We run multiple (10+) sequencers (various HiSeq, MiSeq) with a back-end infrastructure that handles the data. Doing processing offline gives us a unified way of managing data and streamlines things in general.

[WhatsOEver]

We run multiple (10+) sequencers (various HiSeq, MiSeq) with a back-end infrastructure that handles the data. Doing processing offline gives us a unified way of managing data and streamlines things in general.

Sorry for being stupid, but I still don't get it
We currently only have the MiSeq (other sequencing is outsourced), but we also have a back-end infrastructure processing the data offline. Nevertheless, we only transfer the fastq files after the run is completed and not the whole bunch of data written to disk. How often do you need this additional amount of data? I'm of course interested whether we are discarding useful information here?!

[GenoMax]

Don't think you are discarding useful information.

Every facility has its own way of doing things. We like to process data allowing for mismatches and no mismatches on index reads at the same time. Some of our customers don't want mismatch (tags) data but some do. It is easy to do multiple demux runs all at the same time when you have raw data available off-line. We also keep a backup copy of all raw data (local institutional policy) as a tar archive. It is like insurance. Most of the times you don't need it but when you do it is there. We work on multi-year projects so this is a contractual requirement.

[kcchan]

Like GenoMax says it depends on the setup of the individual facility. For a core lab or service provider it makes sense to keep all the raw data in case customers need to have their data re-processed. MiSeq Reporter requires that the instrument be idle and can take an hour or two to regenerate FASTQs while bcl2fastq can do the same thing in about 10 minutes on our server.