[ronaldrcutler]

Thanks for the insightful info GenoMax. I had always thought reads were in a random order in .fastq files. Then I would also think that reads from two biological replicates where each had 3 sets of PE reads from different lanes would also have a relatively similar order.

In essence R1 and R2 are technical replicates due to bridge amplification. I am going to try and compare any differences I find in read counts using a seed in a sample and not using one. And might even compare alignment between BBMap, Hisat2, and Tophat2

[GenoMax]

Thanks for the insightful info GenoMax. I had always thought reads were in a random order in .fastq files. Then I would also think that reads from two biological replicates where each had 3 sets of PE reads from different lanes would also have a relatively similar order.

Reads include positional information (lane, X, Y location) in the header which is only valid for a specific fragment/cluster. Ultimately the physical position of a cluster does not matter (as far as sequence goes).

In essence R1 and R2 are technical replicates due to bridge amplification. I am going to try and compare any differences I find in read counts using a seed in a sample and not using one. And might even compare alignment between BBMap, Hisat2, and Tophat2

Only if they are fully overlapping though I have never thought of R1/R2 as technical replicates of each other. In most cases R1/R2 would not be expected to completely overlap (there are special applications where they are expected to overlap partially).

You could use reformat.sh from BBMap to sample reads for use with all three programs. It will accept this parameter for reproducible sampling.

sampleseed=-1           Set to a positive number to use that prng seed for sampling (allowing deterministic sampling)

[ronaldrcutler]

Reads include positional information (lane, X, Y location) in the header which is only valid for a specific fragment/cluster. Ultimately the physical position of a cluster does not matter (as far as sequence goes).

Order with respect to the fragment to ensure that R1 and R2 from a specific fragment will be at the same relative location in both files. Sequencing is always in 5'->3' direction.

Just for clarification - If I wanted to look at a fragment that is close to the beginning of the R1 file, the same fragment would be near the end of the R2 file. This preserves relative location so the two reads can match up.

In regards to the physical location of a cluster, it does not matter in the reads because it is the job of the aligner to map it to the transcriptome?

[GenoMax]

No. The order of reads is the same in both R1/R2 files.

R1_file

Cluster_1_R1_read
Cluster_2_R1_read
Cluster_3_R1_read

R2 file

Cluster_1_R2_read
Cluster_2_R2_read
Cluster_3_R2_read

Position of the clusters (which is encoded in the fastq header) is only important for the sequencer during sequencing since instrument software keeps track of each individual cluster (there are millions per lane) over the duration of the run. The position itself has no influence on the sequence of that cluster (as long as the software that does the base-calling has done its job right).

[ronaldrcutler]

Alright, that clears a lot up. Your expertise is appreciated.