使用晶片參考資料庫及簡易學生T檢定之基因表現晶片預測分析

Abstract

收集在網路資料庫公開發表過型號為HGU-133A的艾菲爾基因晶片，挑選無患病的一般正常晶片集合成晶片參考集合，利用R的套裝軟體refPlus預處理新的目標晶片，無須再一次同時預處理晶片參考集合和新目標晶片，並使用bar code的原則挑選一些代表基因和簡易學生T檢定方法建立六種不同的分類方法，比較我們所建立的分類法與PAM分類法之晶片分類能力優劣。

We collect published Affymetrix GeneChip HGU-133A arrays from AE(ArrayExpress) and GEO (Gene Expression Omnibus), and select normal control arrays to build up a reference set. R package RefPlus is used to preprocess new target arrays without re-preprocessing them along with the reference set together again. We pick up some “representative” genes through the idea of bar code and build up six classifiers by simple t tests. We then compare the classification abilities between our six classifiers and PAM.