mRNA Analysis Pipeline

Introduction

The GDC mRNA quantification analysis pipeline measures gene level expression in HT-Seq raw read count, Fragments per Kilobase of transcript per Million mapped reads (FPKM), and FPKM-UQ (upper quartile normalization). These values are generated through this pipeline by first aligning reads to the GRCh38 reference genome and then by quantifying the mapped reads. To facilitate harmonization across samples, all RNA-Seq reads are treated as unstranded during analyses.

Data Processing Steps

RNA-Seq Alignment Workflow

The mRNA Analysis pipeline begins with the Alignment Workflow, which is performed using a two-pass method with STAR. STAR aligns each read group separately and then merges the resulting alignments into one. Following the methods used by the International Cancer Genome Consortium ICGC (github), the two-pass method includes a splice junction detection step, which is used to generate the final alignment. This workflow outputs a genomic BAM file, which contains both aligned and unaligned reads. Quality assessment is performed pre-alignment with FASTQC and post-alignment with Picard Tools.

Files that were processed after Data Release 14 have associated transcriptomic and chimeric alignments in addition to the genomic alignment detailed above. This only applies to aliquots with at least one set of paired-end reads. The chimeric BAM file contains reads that were mapped to different chromosomes or strands (fusion alignments). The genomic alignment files contain chimeric and unaligned reads to facilitate the retrieval of all original reads. The transcriptomic alignment reports aligned reads with transcript coordinates rather than genomic coordinates. The transcriptomic alignment is also sorted differently to facilitate downstream analyses. BAM index file pairing is not supported by this method of sorting, which does not allow for BAM slicing on these alignments. The splice-junction file for these alignments are also available.

RNA-Seq Alignment Command Line Parameters

Note that version numbers may vary in files downloaded from the GDC Portal due to ongoing pipeline development and improvement.

# STAR-2.4.2a

### For users with access to the ICGC pipeline:

python star_align.py \
--genomeDir <star_index_path> \
--FastqFileIn <input_fastq_path> \
--workDir <work_dir> \
--out <output_bam> \
--genomeFastaFiles <reference> \
--runThreadN 8 \
--outFilterMultimapScoreRange 1 \
--outFilterMultimapNmax 20 \
--outFilterMismatchNmax 10 \
--alignIntronMax 500000 \
--alignMatesGapMax 1000000 \
--sjdbScore 2 \
--limitBAMsortRAM 0 \
--alignSJDBoverhangMin 1 \
--genomeLoad NoSharedMemory \
--outFilterMatchNminOverLread 0.33 \
--outFilterScoreMinOverLread 0.33 \
--twopass1readsN -1 \
--sjdbOverhang 100 \
--outSAMstrandField intronMotif \
--outSAMunmapped Within

### For users without access to the ICGC pipeline:

### Step 1: Building the STAR index.*

STAR
--runMode genomeGenerate
--genomeDir <star_index_path>
--genomeFastaFiles <reference>
--sjdbOverhang 100
--sjdbGTFfile <gencode.v22.annotation.gtf>
--runThreadN 8

### Step 2: Alignment 1st Pass.

STAR
--genomeDir <star_index_path>
--readFilesIn <fastq_left_1>,<fastq_left2>,... <fastq_right_1>,<fastq_right_2>,...
--runThreadN <runThreadN>
--outFilterMultimapScoreRange 1
--outFilterMultimapNmax 20
--outFilterMismatchNmax 10
--alignIntronMax 500000
--alignMatesGapMax 1000000
--sjdbScore 2
--alignSJDBoverhangMin 1
--genomeLoad NoSharedMemory
--readFilesCommand <bzcat|cat|zcat>
--outFilterMatchNminOverLread 0.33
--outFilterScoreMinOverLread 0.33
--sjdbOverhang 100
--outSAMstrandField intronMotif
--outSAMtype None
--outSAMmode None

### Step 3: Intermediate Index Generation.

STAR
--runMode genomeGenerate
--genomeDir <output_path>
--genomeFastaFiles <reference>
--sjdbOverhang 100
--runThreadN <runThreadN>
--sjdbFileChrStartEnd <SJ.out.tab from previous step>

### Step 4: Alignment 2nd Pass.

STAR
--genomeDir <output_path from previous step>
--readFilesIn <fastq_left_1>,<fastq_left2>,... <fastq_right_1>,<fastq_right_2>,...
--runThreadN <runThreadN>
--outFilterMultimapScoreRange 1
--outFilterMultimapNmax 20
--outFilterMismatchNmax 10
--alignIntronMax 500000
--alignMatesGapMax 1000000
--sjdbScore 2
--alignSJDBoverhangMin 1
--genomeLoad NoSharedMemory
--limitBAMsortRAM 0
--readFilesCommand <bzcat|cat|zcat>
--outFilterMatchNminOverLread 0.33
--outFilterScoreMinOverLread 0.33
--sjdbOverhang 100
--outSAMstrandField intronMotif
--outSAMattributes NH HI NM MD AS XS
--outSAMunmapped Within
--outSAMtype BAM SortedByCoordinate
--outSAMheaderHD @HD VN:1.4
--outSAMattrRGline <formatted RG line provided by wrapper>

# STAR-2.6.0c

STAR \
--readFilesIn <fastq_files> \
--outSAMattrRGline <read_group_strings> \
--alignIntronMax 1000000 \
--alignIntronMin 20 \
--alignMatesGapMax 1000000 \
--alignSJDBoverhangMin 1 \
--alignSJoverhangMin 8 \
--alignSoftClipAtReferenceEnds Yes \
--chimJunctionOverhangMin 15 \
--chimMainSegmentMultNmax 1 \
--chimOutType Junctions SeparateSAMold WithinBAM SoftClip \
--chimSegmentMin 15 \
--genomeDir <genome_dir> \
--genomeLoad NoSharedMemory \
--limitSjdbInsertNsj 1200000 \
--outFileNamePrefix <output_prefix> \
--outFilterIntronMotifs None \
--outFilterMatchNminOverLread 0.33 \
--outFilterMismatchNmax 999 \
--outFilterMismatchNoverLmax 0.1 \
--outFilterMultimapNmax 20 \
--outFilterScoreMinOverLread 0.33 \
--outFilterType BySJout \
--outSAMattributes NH HI AS nM NM ch \
--outSAMstrandField intronMotif \
--outSAMtype BAM Unsorted \
--outSAMunmapped Within \
--quantMode TranscriptomeSAM GeneCounts \
--readFilesCommand <zcat, etc> \
--runThreadN <threads> \
--twopassMode Basic

mRNA Expression Workflow

Following alignment, BAM files are processed through the RNA Expression Workflow to determine RNA expression levels.

The reads mapped to each gene are enumerated using HT-Seq-Count. Expression values are provided in a tab-delimited format. GENCODE v22 was used for gene annotation.

Files that were processed after Data Release 14 have an additional set of read counts that were produced by STAR during the alignment step.

mRNA Quantification Command Line Parameters

HTSeq-0.6.1p1

htseq-count \
-m intersection-nonempty \
-i gene_id \
-r pos \
-s no \
- gencode.v22.annotation.gtf

htseq-count \
-f bam \
-r name \
-s no \
-a 10 \
-t exon \
-i gene_id \
-m intersection-nonempty \
<input_bam> \
<gtf_file> > <counts_file>

mRNA Expression HT-Seq Normalization

RNA-Seq expression level read counts produced by HT-Seq are normalized using two similar methods: FPKM and FPKM-UQ. Normalized values should be used only within the context of the entire gene set. Users are encouraged to normalize raw read count values if a subset of genes is investigated.

FPKM

The Fragments per Kilobase of transcript per Million mapped reads (FPKM) calculation normalizes read count by dividing it by the gene length and the total number of reads mapped to protein-coding genes.

Upper Quartile FPKM

The upper quartile FPKM (FPKM-UQ) is a modified FPKM calculation in which the total protein-coding read count is replaced by the 75th percentile read count value for the sample.

Calculations

• RC_g: Number of reads mapped to the gene
• RC_pc: Number of reads mapped to all protein-coding genes
• RC_g75: The 75th percentile read count value for genes in the sample
• L: Length of the gene in base pairs; Calculated as the sum of all exons in a gene

Note: The read count is multiplied by a scalar (10⁹) during normalization to account for the kilobase and 'million mapped reads' units.

Examples

Sample 1: Gene A

• Gene length: 3,000 bp
• 1,000 reads mapped to Gene A
• 1,000,000 reads mapped to all protein-coding regions
• Read count in Sample 1 for 75th percentile gene: 2,000

FPKM for Gene A = (1,000)*(10^9)/[(3,000)*(1,000,000)] = 333.33

FPKM-UQ for Gene A = (1,000)*(10^9)/[(3,000)*(2,000)] = 166,666.67

File Access and Availability

To facilitate the use of harmonized data in user-created pipelines, RNA-Seq gene expression is accessible in the GDC Data Portal at several intermediate steps in the pipeline. Below is a description of each type of file available for download in the GDC Data Portal.