Homopolymer run errors, polyA bias

[sulfobus]

Pyrosequencing has problems with determining the correct number of bases in a homopolymer stretch. However, we notice that this problem is much more frequent with polyA than polyT/C/G. How can that be?

I found a paper describing roughly the same thing (but with polyA/T), but without any explanation. Could it be that the fluorescent reporter used for dATP is weaker than for the other bases, making it more difficult to determine peak size? The problem starts already at a three base repeat, giving ~10-20% sequences with runs of lengths two, four, five and even six bases.

[pmiguel]

Pyrosequencing has problems with determining the correct number of bases in a homopolymer stretch. However, we notice that this problem is much more frequent with polyA than polyT/C/G. How can that be?

I found a paper describing roughly the same thing (but with polyA/T), but without any explanation. Could it be that the fluorescent reporter used for dATP is weaker than for the other bases, making it more difficult to determine peak size? The problem starts already at a three base repeat, giving ~10-20% sequences with runs of lengths two, four, five and even six bases.

Pyrosequencing does not use fluors. It detects the pyrophosphate released by the polymerase from dNTPs as they are incorporated into the nascent strand. PPi release is coupled ultimately to luciferase-catylyzed emission of light.

Most likely this is a software bug I would guess. Have you used Roche software to look at the flowgrams of reads with poly A length miscalls? If they look different than poly A/C/G, then I would suspect chemistry. Otherwise, more like a software issue.

--
Phillip

[pmiguel]

I found a paper describing roughly the same thing (but with polyA/T), but without any explanation.

By the way, that paper is from the dawn of next gen sequencing--2006! It was GS-20 data. I'm pretty sure the situation is no where nears as bad these days. What version of Roche software generated your basecalls?

[sulfobus]

Actually, we don't sequence ourselves but our sequence provider didn't know what could cause this bias. I will bring up the suggestion about software bug with them, and ask them to use the latest version of the basecalling software. Thanks Phillip!

[pmiguel]

Actually, we don't sequence ourselves but our sequence provider didn't know what could cause this bias. I will bring up the suggestion about software bug with them, and ask them to use the latest version of the basecalling software. Thanks Phillip!

That said, I'm not sure that newer versions of the software fix this problem. But it is worth a shot.

[hyjkim]

Pyrosequencing uses dATP-α-S rather than standard dATP. This is due to nonspecific activity in luciferase using standard dATP. Polymerase is able to incorporate the modified dATP-α-S wherase luciferase cannot. However, the ability of polymerase to incorporate a dATP-α-S is less than standard dATP.

The problems with homopolymers you are seeing is a known problem with the platform. If homopolymers are a large concern, you should try a different platform for sequencing.

[pmiguel]

Pyrosequencing uses dATP-α-S rather than standard dATP. This is due to nonspecific activity in luciferase using standard dATP. Polymerase is able to incorporate the modified dATP-α-S wherase luciferase cannot. However, the ability of polymerase to incorporate a dATP-α-S is less than standard dATP.

Good to know!

Yes, now that I look, I see a ref for it:

Ronaghi, M., Karamouhamed, S., Pettersson, B., Uhlen, M., and
Nyren, P. (1996) "Real-time DNA sequencing using detection of
pyrophosphate release". Anal. Biochem. 242, 84–89

Also, apparently the S enantiomer of dATP-a-S is the substrate usable by a polymerase. The R stereoisomer inhibits the polymerization. See:

Gharizadeh et al. (2002). "Long-read pyrosequencing using pure 2'-deoxyadenosine-5'-O'-(1-thiotriphosphate) Sp-isomer". Anal Biochem 301: 82.

Any chemists know of conditions that might racemize the pure S isomers dATP-a-S that Roche is presumably utilizing in their reagents? We would want to avoid that.

--
Phillip