Hisat2 vs. bowtie2 - why does bowtie2 mapped better?

[frymor]

Hi all,

I have 358 pairs pair-end samples from human genome. I am looking for the best way to map them to the reference for differential expression analysis.
I have tested the STAR aligner as well as tophat and Hisat2, but somehow they all fail to map.

Than I have tested the bowtie2 algorithm and found out, that it maps a lot better (AFAIK) than all the others.
As the mapping with bowtie2 takes longer than Hisat2 and or STAR, I would like to try and understand first if the results are correct.
I thought, Hisat2 and STAR work not only faster, but suppose to also better.

These are the commands I am using to map my files

hisat2 -p 15 -x /home/yeroslaviz/genomes/Mus_musculus/HISAT2/grcm38/genome -1 700APAAXX_1_1.fastq.gz -2 700APAAXX_1_2.fastq.gz -S 700APAAXX_1_2.hisat2.sam

bowtie2 --very-sensitive -p 15 --un tmp.unmapped -x /home/yeroslaviz/genomes/Homo_sapiens/GRCh38.p3/bowtie2indexed/Hsapiens.GRCh38 -1 700APAAXX_1_1.fastq.gz -2 700APAAXX_1_2.fastq.gz -S 700APAAXX_1_2.sam

and the results i am getting from the mapping:

[B]hisat2[/B]

33778368 reads; of these:
  33778368 (100.00%) were paired; of these:
    32551450 (96.37%) aligned concordantly 0 times
    564637 (1.67%) aligned concordantly exactly 1 time
    662281 (1.96%) aligned concordantly >1 times
    ----
    32551450 pairs aligned concordantly 0 times; of these:
      20823 (0.06%) aligned discordantly 1 time
    ----
    32530627 pairs aligned 0 times concordantly or discordantly; of these:
      65061254 mates make up the pairs; of these:
        62207013 (95.61%) aligned 0 times
        1237583 (1.90%) aligned exactly 1 time
        1616658 (2.48%) aligned >1 times
7.92% overall alignment rate

[B]bowtie2[/B]


33778368 reads; of these:
  33778368 (100.00%) were paired; of these:
    11345742 (33.59%) aligned concordantly 0 times
    16268991 (48.16%) aligned concordantly exactly 1 time
    6163635 (18.25%) aligned concordantly >1 times
    ----
    11345742 pairs aligned concordantly 0 times; of these:
      2699536 (23.79%) aligned discordantly 1 time
    ----
    8646206 pairs aligned 0 times concordantly or discordantly; of these:
      17292412 mates make up the pairs; of these:
        12165533 (70.35%) aligned 0 times
        3229431 (18.68%) aligned exactly 1 time
        1897448 (10.97%) aligned >1 times
81.99% overall alignment rate

Can someone explain to me, why bowtie2 performs better here?


thanks

[dpryan]

You're aligning to human with bowtie2 and mouse with hisat2...

[frymor]

yes I have seen that. I am trying to delete the post until I can recheck the results, but it was too late