Sample A: Cover Page of Thesis, Project, or Dissertation Proposal
Sample A: Cover Page of Thesis, Project, or Dissertation Proposal
Sample A: Cover Page of Thesis, Project, or Dissertation Proposal
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
n<strong>or</strong>mal looking that are adjacent to the tum<strong>or</strong> and 2 adenocarcinoma samples); one <strong>of</strong> the n<strong>or</strong>mal<br />
samples is missing, presumably f<strong>or</strong> high tum<strong>or</strong> content. These sample biopsies were harvested<br />
using microdissection techniques and then snap-frozen [49].<br />
The most demanding test <strong>of</strong> a diagnostic assay is whether it is effective in predicting the<br />
outcomes <strong>of</strong> an experiment not included in the development <strong>of</strong> the diagnostic set. A third,<br />
experimentally comparable, dataset was selected from GEO (http://www.ncbi.nlm.nih.gov/geo/):<br />
it was published as a meta-analysis <strong>of</strong> 5 Stage-I non-small cell lung cancers (NSCLC), accession<br />
number GSE6253 [33], which will be called the ‘Lu’ data. Four <strong>of</strong> these datasets were from<br />
published studies <strong>of</strong> lung cancer, including the <strong>or</strong>iginal Bhattacharjee dataset. The fifth dataset,<br />
also from the Affymetrix Tm HG_U95Av2 platf<strong>or</strong>m, consisted <strong>of</strong> samples from Washington<br />
University- St. Louis and included 36 adenocarcinoma and squamous lung cancer patients, all <strong>of</strong><br />
which were described as being in stage I <strong>of</strong> cancer progression. This fifth dataset was loaded into<br />
the ProbeFATE database system and our probe cleansing and sample cleansing methods were<br />
applied as an automated pipeline. The final BaFL cleansed dataset included 10 adenocarcinoma<br />
and 15 squamous samples, and 5,311 ProbeSets having at least 4 BaFL-validated probes in<br />
common. This dataset will serve f<strong>or</strong> an a pri<strong>or</strong>i probe selection experiment, based upon the BaFL<br />
cleansing <strong>of</strong> the Bhattacharjee adenocarcinoma and squamous cell carcinoma data.<br />
BaFL Pipeline Components<br />
Probe Filtering<br />
The BaFL pipeline can be divided into two filtering categ<strong>or</strong>ies, the first, ‘probe sequence’,<br />
categ<strong>or</strong>y uses only the nucleotide sequence f<strong>or</strong> determining filters, and the second categ<strong>or</strong>y uses a<br />
signal measurement assessment as a filter. The probe sequence filters eliminate probes which<br />
32