For instance, if we want to measure the expression of different genes in response to different doses of a chemical in mice. Before starting the RNA-Seq workflow, planning is essential. This step in the analysis is crucial for good results, as there is often no saving a poorly designed experiment. There are few important considerations during planning, including replicates, batch effects, and confounding. For RNA-Seq experiments, there is generally low technical variation, so we will invest in biological replicates instead as we have a shorter budget of 12 experiments. Because the more biological replicates we have, the better the estimates are for mean expression and variation, leading to a more robust analysis (at least 3). Also, will experiment with different batches to confound the analysis. As much as possible will try to perform experimental steps across all conditions simultaneously. If you cannot avoid batches, distribute the samples from each group into each batch as the environment may also influence the expression. Finally, avoid confounding your experiment with major sources of variation. For example, in this experiment, we have animals of different sexes, so we will avoid all-male mice as control and all-female mice as treatment, as we won’t be able to differentiate the treatment effect from sex.
Biological Samples OR Library Preparation:
After the well-planed out experiment, we can begin with sample preparation. While preparing RNA-Seq libraries, the samples are harvested, the RNA is isolated, and DNA contamination is removed. The rRNA is removed, or mature mRNA is selected by their poly-A tails. Then the RNA is turned into cDNA, Fragmented, size selected, and adapters are added to generate the RNA-Seq libraries to be sequenced.
Sequence Reads:
The sequencing generates millions of nucleotides sequence called reads. The reads corresponding to the ends of the fragments are Sequenced. The sequence of each read is output into FASTQ files. After acquiring the FASTQ files, we can start with the computational analysis.
Quality Control:
The first step in the analysis is to assess the quality of the raw data. At this step, we ensure something doesn’t go wrong at the sequencing facility and explore the data for contamination, such as vector, adapter, or ribosomal.
Splice-aware Mapping to Genome:
The next step is alignment or mapping to the genome to determine the location on the genome where the reads originated. Since mRNA contains only the exons needed to create the proteins when the mRNA is aligned to the genome containing introns, some of the reads will be split across introns. Therefore, tools for aligning reads to the genome need to align across introns for RNA-Seq. The alignment output gives the genome coordinates for where the read most likely originated from in the genome and information about the mapping quality.
Counting Reads Associated with Genes:
FoWe uses command-line tools for the entire process until the workflow, which can handle the large sequencing files and computational demands. Therefore, the analysis will be performed in R using predominantly Bioconductor packages. We can read into R the count matrix using the “read.csv” function command and specifying the file. The gene count matrix is arranged with the samples as columns and gene IDs as rows. The count values represent several reads or fragments aligning to the exons of each gene.
Following alignment, the reads aligning to the exons of each gene are quantified to yield a matrix of gene counts.
Statistical Analysis to Identify differential expressed genes:
Once we have count data, differential expression analysis is performed by comparing the expression of each gene between the specified conditions. The statistical analysis output includes the log2 fold changes of expression between conditions and the adjusted P-values for each gene. Genes that reach a threshold for significance can be a subset to define a list of significant differentially expressed genes.
For Differential expression analysis, the followings are the steps we need to perform.
The steps in the differential expression analysis are displayed in the workflow, separated into quality control and Differential Expression analysis steps. To start, we will take the count matrix containing the raw counts for each sample and perform the quality control step. First, the counts will be normalized to account for differences in library depth. Then, principal component analysis (PCA) and hierarchical clustering using heatmaps will be performed to identify potential sample outliers and major sources of variation in the data. After Quality control, the differential expression analysis will be performed, including modeling counts, shrinking the log2 fold changes, and testing for differential expression.
In the first step of Quality control, we perform normalization of the raw counts to assess sample-level quality control metrics. The raw counts represent the number of reads aligned to each gene and should be proportional to the expression of the RNA in the sample. However, factors other than RNA expression can influence the number of reads aligning to each gene. We can adjust the count data to remove the influence of these factors on the overall counts using normalization models. The main factors often considered during the normalization of count data are library depth, gene length, RNA composition. Differences in library size between samples can lead to many more reads aligned to genes in one sample versus another sample. Therefore, we need to adjust the counts assigned to each gene based on the size of the library to do DE analysis. Another normalization factor often adjusted for is gene length. A longer gene generates a longer transcript, which generates more fragments for sequencing.
Therefore, a longer gene will often have more counts than a shorter gene with the same expression level. Since DE analysis compares the expression level of the same genes between conditions, we do not need to normalize for length. However, if you were to compare the expression level of different genes, you would need to account for the lengths of the genes. When adjusting for library size, the composition of the library is also important. A few highly differentially expressed genes can skew many normalization methods that are not resistant to those outliers. After the normalization, we get normalized counts for library size; we compare the counts between the differential samples. We explore how similar the samples are to each other about gene expression to assess the quality of our experiment.
You must be logged in to post a comment.