本文選題：生物信息學(xué) + 基因表達(dá)數(shù)據(jù)��； 參考：《天津大學(xué)》2013年博士論文

【摘要】：肝細(xì)胞癌(Hepatocellular Carcinoma, HCC)有很高的死亡率，占原發(fā)性肝癌的70-85％，肝細(xì)胞癌作為最常見類型的肝癌和世界第五最頻發(fā)的惡性腫瘤，是造成癌癥相關(guān)死亡的第三大原因。肝細(xì)胞癌是一種惡性腫瘤，如果不及時治療，平均生存時間遠(yuǎn)低于一年。肝癌主要是乙肝炎病毒(HBV)和丙肝病毒(HCV)導(dǎo)致。世界上超過500萬人攜帶HBV或HCV病毒，這些病毒會導(dǎo)致慢性肝炎，肝硬化和肝細(xì)胞癌。 本文基于斯坦福數(shù)據(jù)庫(SMD)中的肝癌基因數(shù)據(jù)進(jìn)行分析，研究成果體現(xiàn)在理論創(chuàng)新和結(jié)果創(chuàng)新兩個方面。其中理論創(chuàng)新體現(xiàn)在概念和算法的創(chuàng)新上。本文提出了5個全新的概念：基因共表達(dá)模塊，核心基因，特征免疫基因，個性靶基因，標(biāo)簽基因。并且對應(yīng)每個概念依次設(shè)計了5個算法：Pearson凝聚算法(PAM)，核心基因篩選算法，三層過濾算法，基因協(xié)同過濾算法，基于標(biāo)簽基因的治療方案分類算法。同時，本文給出了大量定義，例如基因社區(qū)網(wǎng)絡(luò)、基因的模塊度、基因影響力、征免疫基因影響力、懲罰準(zhǔn)確率、治療的評分函數(shù)等。 另外結(jié)果創(chuàng)新主要體現(xiàn)在實驗結(jié)果和治療建議創(chuàng)新上，具體從以下幾個方面分析。 ⑴肝癌基因模塊。首次提出GCN網(wǎng)絡(luò)的概念和構(gòu)建方法，，然后用PAM算法和PCC模塊度得到了極強(qiáng)相關(guān)的13個模塊和強(qiáng)相關(guān)的14個模塊。除了一些常見的模塊外，我們還發(fā)現(xiàn)了一些其他功能模塊：止血模塊S1、纖維蛋白模塊W9、抗終止模塊S8、不死模塊S9、抗生長抑制模塊W6、抗凋亡模塊W12、鐵調(diào)節(jié)模塊S6和金屬模塊S。 ⑵肝癌核心基因。提出了基因影響力GF的概念，并使用核心基因算法找到了15個HCC靶基因, HAMP, RNAHP, MT1H, MT1G, MT1L, AQP4, GPC3,MT1E, VIPR1, DNASE1L3, MT1B等。通過構(gòu)建GCN網(wǎng)絡(luò)，獲得3個核心基因HAMP, MTs, GPC3；并依據(jù)這3個核心基因與銅鐵鋅的關(guān)系，提出了治療方案，我們建議給HCC患者適當(dāng)補(bǔ)鋅。 ⑶特征免疫基因。提出免疫特征基因的概念，并設(shè)計了一個三層的過濾器篩選這種基因。找到了23個HCV肝癌的特征免疫基因，例如：MARVELD2,COPB2, HLA-C, MSTP9, TRD@, EPC1, IGL@, TNFSF10.并把這些基因按功能可分為4類：T細(xì)胞類，B細(xì)胞類，免疫信號類，MHC類。由于HBV肝癌的問題出在免疫能力下降，所以治療時候提高患者的免疫球蛋白，T細(xì)胞或者B細(xì)胞的含量等。HCV肝癌中的問題在于病毒更新過快，并且T細(xì)胞也相對不足，建議從抗原或T細(xì)胞著手治療。 ⑷個性靶基因。主要提出了基因協(xié)同過濾算法(GeneCF)，又稱為TOP-N靶基因推薦。該算法的本質(zhì)是根據(jù)患者對某個基因的興趣度的大小進(jìn)行排序。根據(jù)他對基因的興趣度，為每個患者推薦前N個基因給患者，這就是基因TOP-N個性靶基因推薦。在基于準(zhǔn)確率和覆蓋率的基礎(chǔ)上，提出了懲罰準(zhǔn)確率和提出了懲罰覆蓋率的概念。GeneCF算法最大的優(yōu)點，就是受缺失值的影響非常小，尤其在有大量缺失值的情況下，GeneCF算法的優(yōu)越性就更為顯著。 ⑸標(biāo)簽免疫基因。提出了標(biāo)簽基因的概念，并獲得了兩個標(biāo)簽基因。HAMP和GPC3。這兩個基因，對應(yīng)的治療方案為HMAP治療方案和GPC3治療方案。HAMP治療方案主要是使用鐵螯合劑“去鐵”，通過測試52個HCC患者數(shù)據(jù)，發(fā)現(xiàn)不能使用“去鐵”治療的有46%的患者，而使用HAMP方案治療效果優(yōu)良的占總?cè)藬?shù)的20%。尤其注意，低鐵肝癌患者切忌不能使用“去鐵”治療，否則無疑對患者是雪上加霜。所以，給HCC患者使用去鐵方案需要慎重，不可盲從。
[Abstract]:Hepatocellular Carcinoma (HCC) has a high mortality rate, which accounts for the 70-85% of primary liver cancer. Hepatocellular carcinoma is the most common type of liver cancer and the fifth most frequent malignant tumor in the world. It is the third major cause of cancer related death. It is much less than a year. Liver cancer is mainly caused by HBV and HCV. More than 5 million people in the world carry HBV or HCV virus, which can lead to chronic hepatitis, cirrhosis and hepatocellular carcinoma.
Based on the analysis of the liver cancer gene data in the Standford database (SMD), the research results are embodied in two aspects of theoretical innovation and result innovation. The theoretical innovation is embodied in the innovation of concepts and algorithms. This paper puts forward 5 new concepts: gene co expression module, core gene, characteristic immunization gene, individual target gene, and target gene. 5 algorithms are designed in order for each concept: Pearson aggregation algorithm (PAM), core gene screening algorithm, three layer filtering algorithm, gene collaborative filtering algorithm, and label gene therapy scheme classification algorithm. At the same time, this paper gives a large number of definitions, such as gene community network, gene modularity, gene influence, Immunization gene influence, penalty accuracy, treatment score function and so on.
In addition, the results are mainly reflected in the innovation of experimental results and treatment recommendations.
(1) the liver cancer gene module. First proposed the concept and construction method of GCN network. Then, 13 modules and 14 strongly related modules were obtained with the PAM algorithm and the PCC module degree. Besides some common modules, we also found some other functional modules: the hemostatic module S1, the fibrin module W9, the anti termination module S8, and the undead. Module S9, anti growth inhibition module W6, anti apoptotic module W12, iron regulation module S6 and metal module S.
15 HCC target genes, HAMP, RNAHP, MT1H, MT1G, MT1L, MT1G, MT1L, AQP4, GPC3, MT1E, VIPR1, DNASE1L3, etc. were found by the core gene algorithm, and the relationship between the 3 core genes and copper, iron and zinc was obtained by constructing the network. The treatment plan is proposed, and we recommend appropriate zinc supplementation for HCC patients.
(3) characteristic immunization genes. Propose the concept of immune characteristic genes and design a three layer filter to screen this gene. 23 characteristic immune genes of HCV liver cancer, such as MARVELD2, COPB2, HLA-C, MSTP9, TRD@, EPC1, IGL@, TNFSF10., are divided into 4 categories: T cell class, B cell class, immune signal Class, MHC class. Due to the problem of HBV liver cancer in the immune decline, so the treatment to improve the patient's immunoglobulin, T cell or B cell content of.HCV liver cancer, the problem is that the virus is too fast, and T cells are relatively inadequate, suggested from the antigen or T cells in hand treatment.
GeneCF, also known as the TOP-N target gene recommendation. The essence of the algorithm is based on the size of the patient's interest in a gene. According to his interest in the gene, the N gene is recommended to the patient for each patient. This is the recommendation of the gene TOP-N target gene. Based on the accuracy and coverage rate, the greatest advantage of the concept.GeneCF algorithm, which is the penalty accuracy rate and the penalty coverage rate, is proposed. It is that the effect of the missing value is very small, especially in the case of a large number of missing values, the superiority of the GeneCF algorithm is more significant.
The concept of tagging genes, the concept of tagging genes, and two genes of two labelled genes.HAMP and GPC3. are obtained. The corresponding treatment scheme is the HMAP treatment scheme and the GPC3 therapy.HAMP treatment program mainly using the iron chelating mixture "iron removal". By testing the data of 52 HCC patients, it is found that the "iron removal" treatment can not be used. 46% of the patients, and the use of the HAMP scheme to treat the total number of excellent 20%. especially attention, low iron liver cancer patients should not be able to use "iron removal" treatment, otherwise the patient is undoubtedly a frost. Therefore, the use of iron removal for patients with HCC needs to be careful, not blind.
【學(xué)位授予單位】：天津大學(xué)
【學(xué)位級別】：博士
【學(xué)位授予年份】：2013
【分類號】：R735.7;Q811.4

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 李輝;王金蓮;;基于基因表達(dá)譜的腫瘤預(yù)測模型研究[J];電子學(xué)報;2008年05期

2 李小波;;多步驟降維的腫瘤特征基因選擇方法[J];復(fù)旦學(xué)報(自然科學(xué)版);2008年04期

3 陳浩;李建生;莢衛(wèi)東;王偉;許戈良;馬金良;;肝細(xì)胞癌患者血清ECM1、MMP-9和VEGF水平的變化及其意義[J];實用肝臟病雜志;2010年06期

4 阮曉鋼;王金蓮;;基于腫瘤基因表達(dá)譜的基因功能模塊識別[J];北京工業(yè)大學(xué)學(xué)報;2007年04期

5 劉全金;李穎新;朱云華;阮曉鋼;;基于BP神經(jīng)網(wǎng)絡(luò)的腫瘤特征基因選取[J];計算機(jī)工程與應(yīng)用;2005年34期

6 徐菲菲;苗奪謙;魏萊;;基于模糊粗糙集的腫瘤分類特征基因選取[J];計算機(jī)科學(xué);2009年03期

7 王林;戴冠中;趙煥成;;一種新的評價社區(qū)結(jié)構(gòu)的模塊度研究[J];計算機(jī)工程;2010年14期

8 李穎新;李建更;阮曉鋼;;腫瘤基因表達(dá)譜分類特征基因選取問題及分析方法研究[J];計算機(jī)學(xué)報;2006年02期

9 李宏;易麗君;;特征基因選取及腫瘤亞型識別方法研究[J];計算機(jī)測量與控制;2007年06期

10 阮曉鋼;晁浩;;腫瘤識別過程中特征基因的選取[J];控制工程;2007年04期

本文編號：1917117