Assignment #1

Assignment #1 - Data set selection and initial Processing.

Assignment #1 - Data set selection and initial Processing.

---

Objective: Data exploration on expression dataset

Time estimated: 10 h; taken 12 h;

Date started: 2023-2-10 ; completed: 2023-1-14

---

Select an Expression Data Set

• GEOMetadb to find db: rs <- dbGetQuery(con,sql)

• Selected:

  https://www-ncbi-nlm-nih-gov/geo/query/acc.cgi?acc=GSE104406	Aging Human Hematopoietic Stem Cells Manifest Profound Epigenetic Reprogramming of Enhancers That May Predispose to Leukemia (RNA-Seq of HSCe)

Clean the data

1. Download the data with GEO2R and see the infos

   kable(data.frame(head(Meta(gse))), format = "html")

2. Assess data quality for the control and test conditions

   Issue: Error in cpm.default(abs(raw_dat[, 2:21])) : library sizes should be finite and non-negative Solution: summary(raw_dat) instead of table(raw_dat), found a row of NA, removed.

Apply Normalization

• Method?
  • plots: pre- and after-, fig.align to make side by side
  • Used four types of plots
    • It is notable that from MDS Plot, some differences may be caused by gender, should be noticed in future analysis.

Map to HUGO symbols

• Map rows to HUGO gene symbols

  1. Search for human dataset starting with ENSG

     ensembl <- useMart("ensembl")
     kable(head(datasets[grep(datasets$dataset,
                       pattern = "sapiens"),]),format = "html")
     
     # ENSG and HGNC
     kable(searchAttributes(mart = ensembl, 'ensembl|hgnc')[1:12,] , 
           format="html") %>%
       row_spec(c(1,11), background = "yellow")

  2. unmapped rows:

     1	alignment_not_unique
     10	ENSG00000108264

     • 2 types of missing mapping: either is invalid ensembl(i.e. side notes instead of meaningful data); or the emsembl id AND its corresponding hgnc exists, but not included in our mapping db
     • alignment_noy_unique should be removed before in data cleaning
  • No rows that map to more than one symbol
  • Multiple rows that map to the same symbol: should keep all, because cannot expect 1-on-1 mapping

💡 Conclusion and outlook: Selected and cleaned data, normalized and mapped to HUGO symbols

---