R编程讲解、辅导data留学生程序、讲解R语言编程
- 首页 >> Java编程 Homework Assignment #6
Homework should be turned in by uploading a document to BLACKBOARD (BB). The document should include your name in the heading, and should be named with your last name, first initial (or full name if there are others with your initial), and the number of the assignment, with an extension appropriate to the file type (e.g. Smith.J.6.doc or Smith.J.6.pdf). Please put all parts together in the same document.
WHY IS THIS IMPORTANT? Differential gene analysis is a common technique used in RNA-Seq studies. This homework assignment will familiarize you with standard RNA-Seq analysis and visualization practices. You will learn how to define a model matrix, contrasts, block, remove batches/covariates from the data, identify differentially expressed genes using the limma/voom R package and how to compare and contrast two signatures using the results from a pathway enrichment analysis using the goana R function from the limma R package.
Question
Background: Researchers are studying how molecular subtypes of breast cancer may be useful in planning treatment and developing new therapies. The complex profile of each subtype is determined using their gene signatures. Most studies divide breast cancer into 4 major molecular subtypes: Luminal A, Luminal B, Triple negative/basal-like, and HER2-enriched. There are many other less common molecular subtypes, including claudin-low and molecular apocrine types.
Luminal tumor cells look the most like cells of breast cancers that start in the inner (luminal) cells lining the mammary ducts. Luminal A tumors (30-40% of breast cancers) tend to be estrogen receptor-positive (ER-positive), HER2 receptor-negative (HER2-negative), and tumor stage 1 or 2. Luminal B tumors (10-20% of breast cancers) tend to be ER-positive and may be HER2-negative or HER2-positive. Compared to Luminal A tumors, Luminal B tumors tend to have factors that lead to a poorer prognosis including increased tumor stage, larger tumor size, and positive lymph nodes.
In question 2, we will explore how similar (or dissimilar) the enriched pathways are for Luminal A and Luminal B breast cancers using over-representation analysis (ORA).
WHAT TO DO:
Differential expression analysis for Luminal A or Luminal B breast cancers vs. normal adjacent tissue has already been performed for you. A linear fold change of at least 1.5 in the mean expression of a gene and an adjust p-value of at least 0.05 was used to determine statistical significance. The resulting fitted models (microarray linear model fit class (MArrayLM) objects) were saved in rds formatted files. Download the files (HW6.luma.fitted.model.rds and HW6.lumb.fitted.model.rds) from BB.
1)Load the data from the files using the readRDS function. Create a table for each breast cancer subtype for the differentially expressed genes. Order the gene list using the log2 fold change in expression. Filter for only the genes that exceed a linear fold change of 1.5 (log2(1.5) = 0.585) and have an adjusted p-value less than 0.05.
a.In a table, list how many genes are up and down for each signature.
b.How many of the top five up genes and top five down genes are similar between the two?
2)Using the significant (adjusted p-value of no more than 0.05) differentially expressed genes with at least a 1.5-fold change, run over-representation analysis against the GO database on each dataset. Use the function limma::goana and consider only the biological process ontology (BP). For each dataset, report the top 5 up-regulated pathways and top 5 down-regulated pathways in a table. Reminder: the genes associated with each dataset were already tested for significance using a linear fold change of at least 1.5-fold.
3)Combine the top 50 up pathways from each dataset. Before selecting the top 50 pathways for either dataset, sort the pathways by their statistical significance. Repeat for the top 50 down pathways for each dataset.
a.In a table, report the number of up pathways in common for these two datasets and the number of up pathways unique to each dataset.
b.In a table, report the number of down pathways in common for these two datasets and the number of down pathways unique to each dataset.
4)Save and report the R script.
Give yourself a nice pat on the back – you’ve done a bunch of different kinds of statistical analyses and you should have a nice set of tools at your disposal for the future.
Homework should be turned in by uploading a document to BLACKBOARD (BB). The document should include your name in the heading, and should be named with your last name, first initial (or full name if there are others with your initial), and the number of the assignment, with an extension appropriate to the file type (e.g. Smith.J.6.doc or Smith.J.6.pdf). Please put all parts together in the same document.
WHY IS THIS IMPORTANT? Differential gene analysis is a common technique used in RNA-Seq studies. This homework assignment will familiarize you with standard RNA-Seq analysis and visualization practices. You will learn how to define a model matrix, contrasts, block, remove batches/covariates from the data, identify differentially expressed genes using the limma/voom R package and how to compare and contrast two signatures using the results from a pathway enrichment analysis using the goana R function from the limma R package.
Question
Background: Researchers are studying how molecular subtypes of breast cancer may be useful in planning treatment and developing new therapies. The complex profile of each subtype is determined using their gene signatures. Most studies divide breast cancer into 4 major molecular subtypes: Luminal A, Luminal B, Triple negative/basal-like, and HER2-enriched. There are many other less common molecular subtypes, including claudin-low and molecular apocrine types.
Luminal tumor cells look the most like cells of breast cancers that start in the inner (luminal) cells lining the mammary ducts. Luminal A tumors (30-40% of breast cancers) tend to be estrogen receptor-positive (ER-positive), HER2 receptor-negative (HER2-negative), and tumor stage 1 or 2. Luminal B tumors (10-20% of breast cancers) tend to be ER-positive and may be HER2-negative or HER2-positive. Compared to Luminal A tumors, Luminal B tumors tend to have factors that lead to a poorer prognosis including increased tumor stage, larger tumor size, and positive lymph nodes.
In question 2, we will explore how similar (or dissimilar) the enriched pathways are for Luminal A and Luminal B breast cancers using over-representation analysis (ORA).
WHAT TO DO:
Differential expression analysis for Luminal A or Luminal B breast cancers vs. normal adjacent tissue has already been performed for you. A linear fold change of at least 1.5 in the mean expression of a gene and an adjust p-value of at least 0.05 was used to determine statistical significance. The resulting fitted models (microarray linear model fit class (MArrayLM) objects) were saved in rds formatted files. Download the files (HW6.luma.fitted.model.rds and HW6.lumb.fitted.model.rds) from BB.
1)Load the data from the files using the readRDS function. Create a table for each breast cancer subtype for the differentially expressed genes. Order the gene list using the log2 fold change in expression. Filter for only the genes that exceed a linear fold change of 1.5 (log2(1.5) = 0.585) and have an adjusted p-value less than 0.05.
a.In a table, list how many genes are up and down for each signature.
b.How many of the top five up genes and top five down genes are similar between the two?
2)Using the significant (adjusted p-value of no more than 0.05) differentially expressed genes with at least a 1.5-fold change, run over-representation analysis against the GO database on each dataset. Use the function limma::goana and consider only the biological process ontology (BP). For each dataset, report the top 5 up-regulated pathways and top 5 down-regulated pathways in a table. Reminder: the genes associated with each dataset were already tested for significance using a linear fold change of at least 1.5-fold.
3)Combine the top 50 up pathways from each dataset. Before selecting the top 50 pathways for either dataset, sort the pathways by their statistical significance. Repeat for the top 50 down pathways for each dataset.
a.In a table, report the number of up pathways in common for these two datasets and the number of up pathways unique to each dataset.
b.In a table, report the number of down pathways in common for these two datasets and the number of down pathways unique to each dataset.
4)Save and report the R script.
Give yourself a nice pat on the back – you’ve done a bunch of different kinds of statistical analyses and you should have a nice set of tools at your disposal for the future.