Data QC

Print
Background and Past Practice (prior to 1-Feb-2010)

According to our sequencer manufacturer, there exists up to 10% sporadic instrument-associated failure due to:

  1. random bad/insufficient injection
  2. bubble in the capillary during polymer filling
  3. cross-capillary fluorescence due to the nature of capillary sequencing and the proximity of the capillaries to each other (~5um) in the detection cell of the 3730xl platform

To ensure the highest service quality, the CPOS has been offering complimentary default reruns for sequencing samples with failed or sub-optimal data.  However, when reruns data were reviewed for a period of 4 months, it was found that instrument-associated failure rate was indeed very low 0.38% (Table 1), while most failures were sample-specific instead.

Table 1. Mid-Aug to Mid-Dec 2009 (4 months data) 

# of samples ran # of default reruns # of rescued % rescued
18,464 617 72 0.38%

We define the rescued criteria as:

  1. CRL improved > or equal 30%  with an absolute increase of 50bp, OR
  2. Overall TS improved at least 5, with a value at least 35

CRL: Contiguous Read Length is the longest uninterrupted stretch of bases with quality higher than 20.
In the evaluation of the quality of each base, not only the quality value of individual base is used, but also those of adjacent bases within a window size of 20.

TS: Trace Score is the average basecall quality value of bases in the post-trim sequence.


Current Revised Practice (effective 1-Feb-2010)

To better allocate the CPOS resources for more efficient service, the CPOS will no longer provide default rerun. 
Rerun will be done as per client's request with charging scheme below:

  Rescued NOT Rescused
original run $0 normal charge
rerun $0 $20

We ask our clients to review their own data and exercise discretion before requesting reruns.
The "Troubleshooting" guide will offer assistance in making an informed decision. 
Clients are also welcome to consult our platform specialists on difficult cases.

Green highlighted cases below are examples of rescued runs.

  improvement in %
original CRL 10% 20% 30% 40% 50% 60%
50 55 60 65 70 75 80
100 110 120 130 140 150 160
200 220 240 260 280 300 320
300 330 360 390 420 450 480
400 440 480 520 560 600 640
500 550 600 650 700 750 800
600 660 720 780 840 900 960
700 770 840 910 980 1050 1120
800 880 960 1040 1120 1200 1280

The CPOS reserves the rights for final discretion over borderline cases.