Chip-seq mapping (using bowtie2 & tophat2) What's different ?

Question

[leenaehyeon]

I mapped my sample (Chip-seq) using bowtie2 & tophat2 to find out about the difference.

1. bowtie2 -x mm10 -1 sample1.fastq -2 sample2.fastq -S sample_mapped.sam

Result:

12729154 reads; of these:
12729154 (100.00%) were paired; of these:
2525573 (19.84%) aligned concordantly 0 times
7713508 (60.60%) aligned concordantly exactly 1 time
2490073 (19.56%) aligned concordantly >1 times
----
2525573 pairs aligned concordantly 0 times; of these:
777420 (30.78%) aligned discordantly 1 time
----
1748153 pairs aligned 0 times concordantly or discordantly; of these:
3496306 mates make up the pairs; of these:
1563238 (44.71%) aligned 0 times
1052885 (30.11%) aligned exactly 1 time
880183 (25.17%) aligned >1 times
93.86% overall alignment rate

I understand mapping rate here is 93.86%

2. tophat2 -o tophat_out_sample -G mm10.gtf --transcriptome-index mm10_transcripts --library-type fr-firststrand mm10 sample1_fastq sample2_fastq

Result:

Left reads:
Input : 12729154
Mapped : 10520541 (82.6% of input)
of these: 1021764 ( 9.7%) have multiple alignments (206443 have >20)

Right reads:
Input : 12729154
Mapped : 9983465 (78.4% of input)
of these: 949034 ( 9.5%) have multiple alignments (206223 have >20)
80.5% overall read mapping rate.
Aligned pairs: 9223914
of these: 837455 ( 9.1%) have multiple alignments
586051 ( 6.4%) are discordant alignments
67.9% concordant pair alignment rate.

I understand mapping rate here is 67.9%

If I was wrong, Each outcome is what does mean? And why do tophat2 & bowtie2 have different outcomes? :( I heard that bowtie2 is more normally used for ChIP-seq.

[Martombo]

tophat is useful when you expect to have large gaps in the alignment of your reads to the reference, which is the common situation if the read is coming from cDNA and could therefore bridge an intron.

Answer 1

[h.mon]

For ChIP-seq, you should map your reads against the genome, so bowtie is an appropriate tool.

Tophat is a splice junction mapper for RNA-Seq reads. From your command line, you are mapping against mouse transcriptome, your mapping rate will be lower than the bowtie mapping because you are sequencing across the genome and mapping on transcriptome only.

[leenaehyeon]

Really happy to your reply. Thank you!!!!

I'll try bowtie2 mapping for Chip-seq! :))))