hisat2 --sra-acc with paired reads producing single read output
Hello there,
I am trying to use the --sra-acc function from hisat2 with paired end data. I have installed both hisat2 and the sra-toolkit successfully. Indeed, the mapping works fine but the SAM output shows reads mapped as if they were single reads. My hisat2 command looks like:
hisat2 --no-mixed --no-discordant -x ../ref//hg38/genome --sra-acc <accession> -S output.sam
where <accession> is a single number, and this SRA accession links to both 1.fastq.gz and 2.fastq.gz
Is there any way to tell hisat2 that the accession refers to paired end reads?
Thanks!
Anna
• 4,079 views
•
link
1 answer
For single end SRA data hisat2 gives summary statistics as:
HISAT2 summary stats:
Total reads: 27870948
Aligned 0 time: 11196555 (40.17%)
Aligned 1 time: 15284312 (54.84%)
Aligned >1 times: 1390081 (4.99%)
Overall alignment rate: 59.83%
For paired end SRA data hisat2 gives summary statistics as:
HISAT2 summary stats:
Total pairs: 9113937
Aligned concordantly or discordantly 0 time: 4233427 (46.45%)
Aligned concordantly 1 time: 4493249 (49.30%)
Aligned concordantly >1 times: 351445 (3.86%)
Aligned discordantly 1 time: 35816 (0.39%)
Total unpaired reads: 8466854
Aligned 0 time: 6882022 (81.28%)
Aligned 1 time: 1428012 (16.87%)
Aligned >1 times: 156820 (1.85%)
Overall alignment rate: 62.24%
From these results it can be seen that hisat2 treats paired-end and single end data differently.
• 0 views
•
link
Log in to answer this question.
Paired-end reads are aligned together. Many aligners drop the read designations from read name and may encode that information in SAM flags (83 and 147).
Check if the SRA data really contain paired-end reads. I have seen single-end data marked as paired-end - in that case, it was a faulty upload. But bottom-line is don't trust SRA blindly.
I really don't know about
hisat2 --sra-accoption, but while downloading the data using NCBI-SRA Toolkit I had the same issue. fastq-dump utility of NCBI-SRA Toolkit gave me single fastq file for paired entry. This issue has been solved by--split-filesoption provided byfastq-dump.