[Bioc-sig-seq] ShortRead and calibrated qualities

Martin Morgan mtmorgan at fhcrc.org
Tue Nov 4 21:46:59 CET 2008


Hi Victor --

I cc'd the bioc-sig-sequencing list, in case this is interesting to 
other people.

Victor Ruotti wrote:
> Hello Martin,
> We are using ShortRead and are very happy with it.
> I was just looking at the graph in the QA report "per-cycle quality 
> score" given by ShortRead. I noticed that the quality scores are 
> "calibrated" using the alignment information. How easy it is to plot the 
> same graph using raw qualities scores, i.e. uncalibrated quality scores?
> 
> Is there a way to do this using ShortReads?

There is not automatic way, but here's what you can do.

The uncalibrated scores are in _prb.txt files, and can be read with

 > prb = readPrb(dirPath, regex)

where regex is a regular expression defining which _prb files you want 
to read in to a single 'prb' object. E.g., the first two tiles of lane 1 
of a solexa run

 > prb <- readPrb(sp, "s_1_000[1:2]_prb.txt")
 > prb
class: SFastqQuality
quality:
   A BStringSet instance of length 58018
         width seq
     [1]    36 hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh
     [2]    36 hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh
     [3]    36 hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhSh
     [4]    36 hhhhh]ChUhhhhhhhhhhBhhhhG`Jhh_hhWhMN
     [5]    36 hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhM
     [6]    36 hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh
     [7]    36 hhhhhhhhhhhhhhhhhhhhhRhhhhhhhhhMXhSA
     [8]    36 hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh
     [9]    36 hhhEhhhhhhhahhhhhOhEIAAKChhhhhhhUGhH
     ...   ... ...
[58010]    36 hhhhhhhhhhPhch`hhRQhWKCGP?BGPIDPSNJL
[58011]    36 GVQhE_E at U`VULMFKIUHPQ>KFH>ACGGH??BH=
[58012]    36 JNeSPWTJI]NUPHIEJMMEHHTECJNIHIKFB>HA
[58013]    36 I;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
[58014]    36 IGJLEG?GO@?AJ?L>??KIBBIBAI?D?>=F@?BE
[58015]    36 hheJShSh_cThhh[ghh\MZKQJI@[SPLC at REJJ
[58016]    36 Bhc^eAR>a>hhLY@=EDDCCVEJM=HT?@D>BF>R
[58017]    36 Shh at hFQhhKhhh`ZJhhhh]hbRh[hfh]dIhhAh
[58018]    36 [hhhhhhh[hhhhhh`hghNbh[=hd?I`hheHhTP

 From here you can, for instance,

 > m = as(prb, "matrix")
 > colMeans(m)

 > dim(m)
[1] 58018    36
 > colMeans(m)
  [1] 36.68253 34.77074 34.76543 33.89381 33.70597 33.06138 32.16550 
32.02453
  [9] 32.00359 31.28284 31.17655 30.73650 30.98292 30.23918 29.65099 
28.58637
[17] 27.54683 27.70426 27.19865 25.85327 25.37835 24.94302 24.83817 23.99030
[25] 23.41432 23.42025 22.13992 21.79825 20.64564 19.83484 19.90560 18.64320
[33] 17.63711 17.35210 16.81506 16.62260

Martin

> Thanks in advance.
> Victor Ruotti
> 


-- 
Martin Morgan
Computational Biology / Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.
PO Box 19024 Seattle, WA 98109

Location: Arnold Building M2 B169
Phone: (206) 667-2793



More information about the Bioc-sig-sequencing mailing list