read length of GS FLX Titanium

[anabio]

So far, the average read length of 454 GS FLX Titanium system in our lab is 400 base pair, with most less than 400 base pair. Lately, a run got around 320 base pair with less homopolymers in the sequenced genome. The library prep is ok. We don't know whether it is common for 454 GS FLX Titanium, or there is something else we can improve.

How about your read length?

[westerman]

With us running different projects from different PI with different library preps it is hard to break out any good statistics. But I would say that if you have been averaging 400 bp reads then you are doing good. Ours tend to be worse but then as I said -- multiple projects in any given run. Your run with 320 bp ... well ... I would be tempted to just chalk it up to a bad day unless subsequent runs are equally poor.

As for our stats, here are the last 3 runs off of the machine. Take the numbers as you will. Some of the preps were fairly bad.

Total Number of Reads Basecalled:,610840
Average Sequence Length:,340.105
Standard Deviation:,131.506

Total Number of Reads Basecalled:,1170569
Average Sequence Length:,385.581
Standard Deviation:,122.721

Total Number of Reads Basecalled:,461593
Average Sequence Length:,290.199
Standard Deviation:,161.673

[anabio]

400~500bp is expected with 200 cycles. According to westerman's stats, the standard Deviation is considerable. I think there must be a QC limiting the read length to ensure high quality data. After cycles, the signal from certain wells were too low to be recorded because of nonsynchronous extension or weak signal from lost copies during wash. If this is the truth, how to improve？If not...Could anyone got longer read length give some suggestion？

[westerman]

I do want to repeat that our runs tends to be mixed samples. Thus it becomes hard to troubleshoot problems since we are not running the same stuff over and over again. For example that last run had five samples in four regions (two samples were barcoded together). Looking at the per-region stats shows [length, std. dev]:

region 1:,259.646,169.349
region 2:,328.566,146.516
region 3:,62.784,39.604
region 4:,280.595,156.995

The controls in all 4 regions were mostly the same with about 330 bases, s.d., 146 bases.

So obviously the second sample was great (i.e., ran as good as the controls) while the third sample had prep problems.

Our second to the last run we did much better both for the samples and for the overall run.

region 1: 389.056,125.754
region 2: 380.148,129.73

With the control close to 445 bases, sd of 90.

-------------

Going back to anabio's original message, what library and controls stats do you normally get?

[pmiguel]

I do want to repeat that our runs tends to be mixed samples. Thus it becomes hard to troubleshoot problems since we are not running the same stuff over and over again. For example that last run had five samples in four regions (two samples were barcoded together). Looking at the per-region stats shows [length, std. dev]:

region 1:,259.646,169.349
region 2:,328.566,146.516
region 3:,62.784,39.604
region 4:,280.595,156.995

The controls in all 4 regions were mostly the same with about 330 bases, s.d., 146 bases.

So obviously the second sample was great (i.e., ran as good as the controls) while the third sample had prep problems.

Our second to the last run we did much better both for the samples and for the overall run.

region 1: 389.056,125.754
region 2: 380.148,129.73

With the control close to 445 bases, sd of 90.

-------------

Going back to anabio's original message, what library and controls stats do you normally get?

What Rick isn't telling you is that these samples are all cDNAs using the SMART kit. With which we frequently see a bimodal read length distribution with a tail extending towards 40 bases. I'm hoping all these issues go away with the Roche cDNA protocol.

The Titanium controls comprise several template lengths, I believe. Might want to focus on the long ones only if you want to troubleshoot read length.

Even if you do that, it is possible that the issue I'm going to call "polyA glare" of nearby sample beads could negatively impact the read length of the control beads.

--
Phillip