流感病毒蛋白質序列之特徵搜尋與變化分析

Abstract

由於A型流行性感冒的高變異性提高人類應對流行性感冒的難度，因此難以預測未來可能發生的流感病毒突變，學者使用不同的方法找出流感病毒蛋白質序列中具有特別意義用以判別相異物種的胺基酸位置，將這些特別的胺基酸位置稱之為重要特徵，透過提出的重要特徵探討於流感病毒蛋白質序列中其胺基酸突變演化可能的代表意義。本研究利用ARI(Adjusted Rand Index)計算物種間流感病毒蛋白質序列不同胺基酸位置成為重要特徵的可能性，其ARI值作為尋找重要特徵之參考相對值，本研究依照ARI值由大到小排序僅提出前20名特徵位置(Top 20)，針對不同年代進行搜尋重要特徵討論，分析不同物種之流感病毒蛋白質序列隨著年代變遷其重要特徵之變化，藉由Top 20重要特徵位置於不同年代間的胺基酸殘基變化，提供相關研究人員推測分析流感病毒突變演化趨勢，用以評估預測未來發生大流行之可能性。

The high mutability of influenza A virus makes the prediction difficult, which has drawn a lot of attention from researchers to cope with epidemic and pandemic influenza. Among many topics is the finding of sequence signatures in influenza virus. Unlike previous works, we propose to apply adjusted Rand index (ARI) as a measure to identify potential signatures. According to the species and the period, we calculate the ARI value of each position in the influenza protein sequences, and output the positions with the top-20 ARI values as the candidate signatures. We compare our results against those in previous computational studies and also in biological experiments. The analysis shows that our new approach can discover more biologically verified signatures than were identified by other computational methods. Furthermore, our investigation of the change of signatures over different periods of time provides the clues and hypotheses of possible virus mutations.