4.2 Computational Analysis

The RNA-DNA approximity is aquired by chimeric reads which contains RNA information at one end and DNA at the other. To understand the raw data right after sequencing, next we'll show the basic analysis pipeline.

Both MARGI and GRID-seq use pair-end strategy, so they follow very similar pipeline. However, CHAR-seq uses single-end sequencing, it differs a little in some details.

Pipeline Reference:

Pipeline

Trim artificial bases:

Different linker design may contain some artificial sequence. for MARGI protocol, the two random bases we have to remove of RNA end 5'.

Trim adaptors:

Removal of adapter sequences in a process called read trimming, or clipping, is one of the first steps in analyzing NGS data, before mapping, people have to trim the sequencing adaptors which were added for through library preparation. These adaptors were used for holding barcoding sequences, forward/reverse primers (for paired-end sequencing) and the important binding sequences for immobilizing the fragments to the flowcell and allowing bridge-amplification.

There are a bunch of tools in your toolkit to remove adaptors, in MARGI portal FastX is used.

Remove duplicates:

This step also follows the classic NGS analysis. The main purpose of removing duplicates is to mitigate the effects of PCR amplification bias introduced during library construction. Furthermore you gain the computational benefit of reducing the number of reads to be processed in downstream steps.

Tools for removing duplicates including Picard and Samtools and so on, in MARGI portal fastuniq is used.

Separate RNA and DNA end:

One of the key steps across all different protocols is that we have to separate RNA and DNA from the chimeric sequence. For pair-end sequencing like MARGI, the information has been identified though linker design, which naturally separates RNA and DNA information into two reads. However, with Single-end sequencing, CHAR-seq done this by finding the bridge signal in the sequence.

CHAR-seq: RNA and DNA portions of each read are identified and split by a custom python script that finds the bridge, uses the bridge polarity to identify which side of the read is RNA or DNA, and verifies that there is only a copy of the bridge. All reads lacking a full bridge sequence or that contained two or more bridges were removed. Only reads that contained sequence on both sides of the bridge (RNA and DNA) were processed further.

GRID-seq: To precisely remove the linker sequence from each read, MmeI motifs were used for defining linker boundaries. Linker orientation also dictated whether a given read at each end was originated from RNA or genomic DNA. raw reads clipped at MmeI motifs to produce paired DNA and RNA read mates ranging from 18 bp to 23 bp.

Map to the reference:

Not like traditional mapping strategy that feed the pair-end reads (usually reads1.fastq, reads2.fastq) into the aligner simultaneously, RNA and DNA ends are usually treated differently. One consideration is that, the RNA we could detect is the post-processing product after possible splicing and recombination (also called alternative splicing). However, DNA end should be aligned without large fragments insertion.

MARGI:

  • Use splicing aware aligner STAR.

  • It allowed the aligner to clip bases at their 3’ end. because we don't know where the exact meet of RNA and DNA, so align through two ends is a secure strategy.

  • Removing the two random bases attached to their 5’ end.

  • After mapping the RNA and DNA reads, the corresponding mates were merged into a single BAM file using a python script (fixBAMmates).

  • Extra step for identifying diRNA (defination could be found in previous chapter) and pxRNA.

CHAR-seq:

  • RNA and DNA portions of each read are identified and split by a custom python script that finds the bridge, uses the bridge polarity to identify which side of the read is RNA or DNA.

  • Valid sequence only allows one bridge.

  • Unique read IDs from the initial read containing the full molecule remained linked to both the RNA and DNA sequences after splitting from the bridge sequence.

  • Use bowtie2 as aligner.

  • Forcing sense strandedness as bridge orientation should preserve the RNA strand information.

  • Reads that aligned with equal alignment score to more than one transcriptome were filtered based on a priority rank: tRNA, miscRNA, ncRNA, transcript, three_prime_UTR, five_prime_UTR, exon, intron, miRNA, gene, gene_extended2000.

GRID-seq:

  • Bowtie2 with parameter of–local.

  • short reads mapping may introduce non-specificity.

Generating RNA-DNA contact information

Last updated