RNA-SEQ QUANTIFICATION
RNA-seq alignment
1. RNA-seq alignment methodology
• BWA is used for DNA alignment, not for RNA -seq data alignment.
• It is preferred to use splice-aware aligners such as TopHat (BW), HISAT (BW), or STAR (Suffix array).
Standard alignment
• The reads represent RNA-seq fragments aGempHng to align directly to the genome.
• In this case, the aligner is not aware of splice juncHons, so it only aligns reads that perfectly match conHguous
regions of the genome (without interrupHons).
• This creates issues with reads covering exon juncHons, as they do not fully match the genome due to the
presence of introns.
• In spliced genes (with introns), many reads fail to align correctly to the genome, leading to lost or misaligned
reads.
Splice-aware alignment
• Splice-aware aligners (e.g., STAR, HISAT2, or TopHat) consider exon juncHons in spliced genes.
• In this case, reads spanning exon juncHons (represented by dashed lines) are split and aligned correctly to the
genome, even if the sequence includes an intron.
• This approach allows reads to be mapped both within exons and across splice juncHons, accurately reflecHng the
structure of the transcribed gene.
2. Splice-Aware Alignment
• Need genome index file. Transcript annotaHon file is opHonal, but not required.
• For longer reads, map shorter segments (around 25 bp). For normal reads map using around 50bp.
• General steps:
§ Map to exons first.
§ Create juncHon database.
§ Map unmapped reads to juncHons.
3. BAM file for pair-end RNA-seq alignment from STAR
• Column 1: Read ID.
• Column 2: binary encoding flags such as the reads were single- end (0) or pair-end (1). In addiHon, all the pair-
end samples have an odd number (if these number appears directly, you know it´s pair-end, if it’s single-end a 0
would be there; 163 means first lecture of a pair-end (forward lecture) and 83 means the second lecture of a
pair-end (reverse lecture).
• Column 3: Chromosome.
• Column 4: Beginning of the locaHon of the read.
• Column 5: MAPQ (mapping quality), someHmes it`s not calculated (255 indicates that).
The Phred (MAPQ= -10 . log10 (probability of a incorrect alignment)) is used to calculate MAPQ and the values
goes between 0 to 60 (60 would be perfect alignment, 30-60 a reliable alignment and <30 is not a reliable
alignment). You can select certain reads that are above a MAPQ threshold.
• Column 6: Cigar string. The number and posiHon of the mismatches inside a read that has been mapped
towards a certain genome region of reference. For example, in the first read it tells you that the first 27
nucleoHdes were matched (27M) and there is a gap of 1099bp intron (1099N), then the followed 48 matched in
the next exon (48M).
• Column 7: RNEXT, is = since the other pair-end read is in the same chromosome that the RNAME (this read).
• Column 8 and 9: the locaHon of its mate and how long the two mates spanned (covered) on the reference
genome.
• Read sequence.
• ASCII quality control of each nucleoHde sequenced.
• OpHonal:
§ NH: how many places this read is aligned to.
§ nM: number of mismatches for the pair of reads.
§ XS: strand of the underlying transcript generaHng the read for HISAT.
o XA:A: - à in the DNA the + strand was copied and the – was de coding strand.
o XA:A: + à in the DNA the -strand was copied and the + was de coding strand.