find error rate in Raw Read

[mido1951]

Hi,
How can i find the overall error rate of the Raw reads? and the rate of (substitutions, deletions and insertions)?
how to do?
Thanks

[GenoMax]

Look at @Brian's post (#18) here. You would either need a reference or you will have to assemble your data into a reference.

[mido1951]

i used BBmap for mapping. it is a good tool?
this is my result:

   ------------------   Results   ------------------

Genome:                 1
Key Length:             13
Max Indel:              16000
Minimum Score Ratio:    0.56
Mapping Mode:           normal
Reads Used:             108490  (52257243 bases)

Mapping:                4967.598 seconds.
Reads/sec:              21.84
kBases/sec:             10.52


Read 1 data:            pct reads       num reads       pct bases          num bases

mapped:                  29.4433%           31943        29.9910%           15672446
unambiguous:             29.0607%           31528        29.6165%           15476766
ambiguous:                0.3825%             415         0.3745%             195680
low-Q discards:           0.0000%               0         0.0000%                  0

perfect best site:        0.0046%               5         0.0002%                126
semiperfect site:         0.0046%               5         0.0002%                126

Match Rate:                   NA               NA        85.4575%           14050523
Error Rate:              40.9037%           31938        14.5421%            2390951
Sub Rate:                40.9024%           31937         7.2366%            1189814
Del Rate:                40.7756%           31838         4.6777%             769083
Ins Rate:                40.7103%           31787         2.6278%             432054
N Rate:                   0.0013%               1         0.0003%                 55

Total time:             5126.959 seconds.

in this example: the global error rate is 40.9024% or 14.5421%(per bases)?
and for Sub, Del, Ins?
thanks

[GenoMax]

BBMap is a great tool but it needs to be applied appropriately.

You need to provide additional information to get more. What kind of data is this, what are you mapping against, the command line options you used. BTW: You have only 30% of the reads mapping so that is low numbers to begin with (if you are mapping against a reference).

[mido1951]

yes, I use my raw Reads for mapping to the reference.
and i use the default parameters for BBmap.
in this example: the global error rate is 40.9024% or 14.5421%(per bases)?
and for Sub, Del, Ins?
thanks

[GenoMax]

It appears that you did not read the post #18 in the thread that I linked above that tells how to plot the rates you are looking for.

If you are working with MinION or PacBio type data (long reads) then you should be using a mapPacBio.sh instead of bbmap.sh. That error rate may not be meaningful as it stands now.

To calculate the error rate (for long reads) you may have to do something like this:

$ mapPacBio.sh in=your_reads.fa ref=ref.fa mhist=mhist.txt qhist=qhist.txt maxlen=2000

@Brian is likely to swing by this thread later tonight (and may have specific suggestions). That example above was for PacBio data but I assume it may work for MinION data.

[Brian Bushnell]

Well, I'm not really sure what those reads are, at an average length of ~481bp. Probably PacBio, though, considering the ~14.5% error rate.

As GenoMax said, you should map PacBio (or minIon) reads with mapPacBio.sh. The usage and algorithm are the same as bbmap.sh, but it is designed for the PacBio error model.

The error rates you want are in the "pct bases" column.

[mido1951]

i work with MinION reads with an average lenght ~5000bp.
For this i use mapPacBio.sh?

excuse me, for the error rate i work with "pct bases" column?
thanks

[Brian Bushnell]

i work with MinION reads with an average lenght ~5000bp.
For this i use mapPacBio.sh?

Yes. And you may need to use higher-than-default sensitivity, if the data is particularly low quality; you can adjust sensitivity with the "minid" flag. E.g. "minid=0.5" will try to map reads down to 50% identity.

excuse me, for the error rate i work with "pct bases" column?
thanks

That's correct.

[mido1951]

I used mapPacBio.sh and mapPacBio8k.sh.
but the execution takes a long time compared to bbmap.sh.
that's logic?

[Brian Bushnell]

Yep, mapPacBio is slower, because it supports higher sensitivity and longer reads. Note that "mapPacBio8k.sh" is not in the latest release, so you may be using an older version that might also be somewhat slower.