It's more likely that the SRA archive was processed incorrectly, which happens a lot. But typically, you should be able to tell whether the reads were merged based on the quality profile and read lengths. Read lengths will typically fall in some kind of bell-curve distribution if the reads were merged, and the quality scores should abruptly increase in the middle of the read and be otherwise roughly symmetrical. I think it would be unusual (and a bad idea) for people to merge reads prior to submission.
Can you print the SRA command you used to decompress the archive? With the right command, an archive that was made from 2 files should produce 2 files, or interleaved reads.
you may use prefetch from SRA toolkit first (less download if you want to try different parameters and tools in SRA toolkit.
I downloaded files. They are normal paired-end files with the same number of reads. and used FastQC. Quality distribution plots look as they should. My guess is that when you downloaded the files last time it terminated prematurely giving you corrupted data.
Do you mean that after using fastq-dump from https://ncbi.github.io/sra-tools/fastq-dump.html with option --split-files and --filter-technical you get one file for paired read experiment? You should get two files with I1 and I2 appended to the name for a paired end submitionCould you please provide SRA ID of this archive? After downloading use FastQC and look at the per position quality. It has to have a distribution where begining and end of read apears of a bit lower quality and with greater standart deviation. Also for the forward read (I2 in file name) quality usualy is of less quality than for the forward (I1 in file name).
You are welcome.
Please use ADD COMMENT to keep things organized when you are not giving an answer to the topic.
Also, there are the bookmark, upvote and accept buttons available to you, so you can close the question and others will know that you found the answer.
Thank you
It's more likely that the SRA archive was processed incorrectly, which happens a lot. But typically, you should be able to tell whether the reads were merged based on the quality profile and read lengths. Read lengths will typically fall in some kind of bell-curve distribution if the reads were merged, and the quality scores should abruptly increase in the middle of the read and be otherwise roughly symmetrical. I think it would be unusual (and a bad idea) for people to merge reads prior to submission.
Can you print the SRA command you used to decompress the archive? With the right command, an archive that was made from 2 files should produce 2 files, or interleaved reads.