Comparison: Manual Sequencing vs. Automated Sequencing

Should you hand-sequence your DNA, or should you send it to the U of M Sequencing Core and let them do it?

This is a document written in the late 1990's, when there were still a few relic labs
doing their own manual sequencing. I believe even those modern-day Luddites have
now come to realize the benefits of automated sequencing. This document is kept
here (and even updated occasionally) primarily for historical reference and interest.

What does manual sequencing cost you?

The typical cost for manual sequencing is about $15.71 per sample, broken down as follows: (assumptions: six samples, double-loaded for longest possible read, requires 48 lanes; costs based on data from a lab at U-M performing manual sequencing as of Jan 2001; assume the tech makes $20,000 per year salary.)

  1. Sequenase kit: $2.45/sample
  2. S-35 label: $4.48/sample
  3. Gel components: $1.75/sample (urea, acryl, film)
  4. Misc (tips, buffer, reagents) $0.36 (just to make it all a round number)
  5. 4 hours of *hands-on* time with reactions, pouring, loading, cleanup and gel reading: $40, or about $6.67 per sample.

Bottom line: For each sample, you pay about $15.71, which gets you maybe 300 nt of sequence from each template, hand-typed into your computer (with all the associated errors of hand-entry) and poor capability for subsequent sequence assembly. This cost goes up radically if your gels occasionally fail, if you are running just one or two samples, if you don't use the entire batch of radioactive nucleotides before it decays, or if you have more expensive labor costs. Your cost for radioactive disposal may add *considerable* cost as well.

Our sequencing currently costs $3.00/lane in single-sample quantities, including labor, reagents, sequencer time and support informatics. Just the reagents alone cost almost that for manual sequencing! We get MUCH longer read lengths, automatically entered into a computer (no hand data-entry), ready for automatic assembly. If the project is a bit larger, our price goes down considerably, too!

*** By the way, the lab PI should never be sequencing! *** This is not at all cost-effective.

Are you planning to assemble shotgun sequencing reads?

Here's a HUGE consideration. The programs for assembling shotgun sequence fragments from *autosequencing chromatograms* are superb. Not only are the alignments far easier than anything you could do with manual sequencing, you can resolve discrepancies swiftly and accurately ... but ONLY with autosequencing chromatograms. Nobody in their right mind would assemble sequence manually these days. Do yourself a favor - find out how easy sequence assembly is with automated sequencing data and modern assembly programs (for example, try Sequencher, or LaserGene or the phred/phrap/consed suite of programs).

How many samples are you sequencing?

Yeah, the Core is occasionally backlogged. Even with that, there are speed advantages (and with our new high-throughput sequencer, turnaround times are typically 48 hours now).

On typical manual gels you can do only 6 samples at once, and it will take you 2 elapsed days - or more, for longer exposures - to sequence them by hand, assuming you have no problems. Weigh that against the Core's capacity: in that same amount of time, you can get dozens or even hundreds of samples completed. Even if we are backlogged and you have to wait for 4 days (check the current backlog), you can get ten times as many samples through the Core than you are likely to do by hand.

How long are your read lengths?

Unless you're running 'King Kong' gels, your manual gels are probably getting 200-300 nt per lane. We typically get at least 800 nt and usually more (at least from a good template - that's a key point). That saves you time, reduces primer walking and cuts in half the number of shotgun clones you need to pick.

Do you make mistakes when manually reading autoradiograms?

Of course you do; human reading of gels is unavoidably error-prone. Automated sequencers eliminate the steps of manual gel reading and entering the sequence into the computer. Not only is it much easier, but it's much less error-prone as well. Yes, autosequencers make basecalling mistakes, but with practice (or with the right assembly software), it is simple to spot and correct those. Besides, do you really enjoy hand-reading gels??

Are you eligible for discounts with the Sequencing Core?

If you don't have to pay full price for sequencing, it's an even better deal. Besides the bulk discounts mentioned above, there are discounts for participants of various Research Centers and sometimes certain departments. It's a bit like receiving a mini-grant when a Research Center gives you access to several thousand dollars in discounted sequencing! See the page Cost of Automated Sequencing for a complete listing of current discounts.

How much time do you have to waste sequencing or training someone to sequence? How *often* do you want to spend that time?

Sure, you may have a grad student or post-doc who knows how to sequence, but what happens when that person leaves? Will they train someone before they go? Can you ensure the new person will overlap with the old? Will the PI have to train someone? It costs you a LOT to train someone in sequencing, and it's not a one-time investment. The Core, however, has lots of technicians, so there's a pool of trained people and new ones can easily be added. Besides, your grad student has better things to do - your research.

On a related note, click here if you are thinking of buying your own sequencer.

Here's the bottom line:

  • Automatic sequencing ...
    • is far less error-prone than manual sequencing,
    • makes assembly WORLDS easier,
    • is almost always cheaper than manual sequencing,
    • produces more data than manual gels,
    • can process more samples in a short period than you could possibly do manually.

Do yourself a favor - let the Core do it.