MiR-PathOgen: 以拓璞資訊方式分析基因與微核醣核酸調控之生物途徑

Abstract

藉由分析微陣列生物晶片產出的高通量基因體資料，探討生物體內複雜的基因調控機制與癌症機轉，已成為現今基因體醫學研究的主流。由於在生物途徑層次上進行分析較具有生物意義，目前的研究逐漸轉往此方向分析。此外，因為上游基因在生物途徑中的影響層面比在下游的基因來得重要，若能結合拓璞結構訊息，可能可以得到更具有生物意義的結果；然而，此新興方向仍在初始發展階段。另一方面，近年來的研究已知微核醣核酸廣泛地在生物體中扮演關鍵角色，它會藉由抑制其目標基因的表現以調節細胞功能。雖然微核醣核酸與其目標基因的預測資料庫已為數不少，但是結合微核醣核酸進行生物途徑分析的方法卻仍十分缺乏，至今尚無文獻發表將微核醣核酸與其目標基因的交互作用直接加入生物途徑共同進行分析的方法。因此，我們希望可以將微核醣核酸和基因的表現變化值結合拓璞結構訊息，在系統的層面進行生物途徑分析，進一步瞭解癌症在生物體內的運作機制。 於此篇碩士論文中，我們提出一個新的方法，將微核醣核酸的調節加入生物途徑中，新建一個包含微核醣核酸的生物途徑資料庫，進行基因體資料分析。此方法主要由三個演算法組成，分別為過度表現分析、影響因子分析，以及系統信任度三種方法，使用MATLABR軟體建構，並命名為MiR-PathOgen。我們首先收集來自PID和BioCarta的生物途徑資訊，加入來自miRecords 和TarBase經過生物驗證的微核醣核酸與其調控基因的資訊，重建結合微核醣核酸的生物途徑資料庫。接著將有差異表現的微核醣核酸與基因以及它們的表現值輸入資料庫後，採用超幾何分布進行過度表現分析和應用拓璞結構與表現值進行影響因子分析，最後再將兩者結果使用系統信任度結合，得到一個整體分數用來作為排名依歸，進而找出顯著相關的生物途徑。 在此研究中，影響因子分析的模擬結果證實此方法可以挑選出受到較上游的微核醣核酸或基因調控與較有訊息傳遞連結的生物途徑；系統信任度顯示此結合方法可以同時考慮每一條生物途徑在過度表現分析與影響因子分析下的結果。分析實際癌症病患微陣列晶片資料的結果顯示我們的方法可以找到受到一群微核醣核酸與基因正向或負向調控的相關生物途徑，而這些生物途徑已於文獻中被證實與該人類癌症的相關性。由於同時考慮了微核醣核酸與基因在給定癌症上扮演的角色，MiR-PathOgen 具有較佳的可信賴性與可再現性，是一個不僅在基因層次同時也在微核醣核酸層次進行基因體資料分析的演算法。我們期望未來MiR-PathOgen可以應用在與癌症相關的生物途徑分析，進一步瞭解生物系統內複雜的基因調控機制與癌症機轉。

Pathway analysis of high-throughput genomic data generated by microarray has been increasingly prevalent in exploring the underlying molecular mechanisms and complex cancer progression of biological systems. Considering that the upstream genes play more important roles than downstream genes do in signaling pathways, utilization of topology information may identify pathways that are more related to conditions being studied, but it is still in early stage and rather incomplete. In addition, even though microRNA has been reported to play a key role in diverse cellular functions by down-regulating its target mRNAs at the post-transcriptional level, only little research in pathway analysis incorporated with microRNA. Therefore, the purpose of this study is to improve the pathway analysis for a better understanding of underlying mechanisms by integrating microRNA and mRNA microarray profiling in a topology-based approach at the systems biology level. An improved approach, named MiR-PathOgen, is developed as a MATLABR-based application. MiR-PathOgen consists of a reconstructed miRNA-integrated pathway database and three algorithms, including an over-representation analysis, an impact factor method and the reliability of a system. To construct a microRNA-integrated pathway database, we first collected pathway information from BioCarta and pathway interaction database, as well as a list of validated microRNA-gene interactions from miRecords and TarBase. Next, over-representation analysis was executed using hypergeometric distribution. The impact factor method incorporated several biological factors, such as regulatory relationship, gene positions and expression changes. Lastly, the results of above analyses were combined by concepts of the reliability of a system to conduct an overall score for pathway rankings. The simulation results of the impact factor analysis illustrated that this topology-based method was able to robustly report the pathways in condition of connected or central distributed microRNAs and genes with higher ranking. Also, the combination of the over-representation analysis and the impact factor analysis at the system level performed a reliable systematical transformation to identify significant pathways. Furthermore, applying our algorithms in three human datasets suggested that MiR-PathOgen is capable of identifying microRNA and gene activated pathways that were related to a given condition being studied. In summary, MiR-PathOgen is a reliable and reproducible pathway analysis tool to concurrently analyze microRNA and mRNA genomic data, which is beneficial for a better understanding of complex biological systems.