BFAST and read error correction (with SAET or similar tool)

[javijevi]

Hi all,

According BFAST paper(s), this tool is capable of managing read errors in color space of SOLiD data.

Does anybody knows if it is also convenient to clean/correct/purge raw data with accuracy enhancement tools (like SAET) prior to use BFAST? Or maybe is it not wise to do it in order to not alter the BFAST behavior?

I've read that it is highly convenient to use SAET (or similar tool) to improve the quality of data prior to use align/assembly tools, but maybe this recommendation is prior to development of BFAST.

Any experience?

Thanks in advance.

[nilshomer]

Hi all,

According BFAST paper(s), this tool is capable of managing read errors in color space of SOLiD data.

Does anybody knows if it is also convenient to clean/correct/purge raw data with accuracy enhancement tools (like SAET) prior to use BFAST? Or maybe is it not wise to do it in order to not alter the BFAST behavior?

I've read that it is highly convenient to use SAET (or similar tool) to improve the quality of data prior to use align/assembly tools, but maybe this recommendation is prior to development of BFAST.

Any experience?

Thanks in advance.

I do not know how SAET works. I would recommend not using SAET and letting BFAST (I am the author) identify color errors versus variants. If you have a recent run with a known reference you can try with and without SAET to assess the difference in color/base error rates after alignment with BFAST. Please PM me if this is the case, since I would interested in the results.

Nils

[nilshomer]

I just looked at an old DH10B run (2008) and found that the the mapping rate (% of bases mapped) went from 55% to 70% but at the cost of an increased mismatch rate and identified color error rate (even at higher mapping qualities). They (ABI) are recommending to only use this with smaller genomes (less than 200Mb). I will still test it out with skepticism.

[javijevi]

I just looked at an old DH10B run (2008) and found that the the mapping rate (% of bases mapped) went from 55% to 70% but at the cost of an increased mismatch rate and identified color error rate (even at higher mapping qualities). They (ABI) are recommending to only use this with smaller genomes (less than 200Mb). I will still test it out with skepticism.

Do you mean SAET improved the percentage of bases mapped by BFAST from 55 to 70%? Or are they using another alignment other than BFAST?

[nilshomer]

Do you mean SAET improved the percentage of bases mapped by BFAST from 55 to 70%? Or are they using another alignment other than BFAST?

I used SAET just now to increase the percentage of bases aligned with BFAST from 55% to 70% at the cost of noise. This was on very old and thus noisy data set so you may get different results. Also, I did not check how this affected detection of polymorphisms (did it correct away the SNPs?). Hence I would not recommend using it until a more conclusive study is performed rather than my one-time anecdotal experience.

"They" (ABI) use mapreads in Bioscope, whereas I use(d) BFAST.