Technical and biological replicates are plan to address two error strata. The first, simply takes into account random variation that arise from the sequencing process, the second must address both, random sampling variation as well as the variation that is not explained by sampling but by factors that cannot be controlled by the researcher, that are not part of the "treatments" but vary anyway, for example genotype of the individual's included, small differences in relevant factor as temperature or other environmental factors... Re-sequencing a library constitute then a technical replicate, but to have a biological replicate (the really relevant one) you need a new extraction of RNA and constructing independent library, which will include both, technical variation as well as non-controlled factors. Making new runs of your libraries will constitute only technical replicates, segregating reads obtained from the same run will be neither, a technical or biological replicate.
I'm undertaking a global transcriptomic analysis, using RNA-sequencing technology. The cDNA fragments generated from the extracted RNA (from my samples) were sequenced in paired-end manner (that is from one end to the other and vice versa). My data was based on the results gotten from this method. Now I want to confirm if the paired-end method could pass for sequencing replicates? (I mean sequencing a cDNA fragment 2-3 times as a replication of scientific analysis?)
If I understand properly your question, if you need a technical replication I think that resequencing the cDNA library could work. Instead, if you need a biological replication it could be better to produce a new cDNA library from another preparation of your RNA sample.
Technical and biological replicates are plan to address two error strata. The first, simply takes into account random variation that arise from the sequencing process, the second must address both, random sampling variation as well as the variation that is not explained by sampling but by factors that cannot be controlled by the researcher, that are not part of the "treatments" but vary anyway, for example genotype of the individual's included, small differences in relevant factor as temperature or other environmental factors... Re-sequencing a library constitute then a technical replicate, but to have a biological replicate (the really relevant one) you need a new extraction of RNA and constructing independent library, which will include both, technical variation as well as non-controlled factors. Making new runs of your libraries will constitute only technical replicates, segregating reads obtained from the same run will be neither, a technical or biological replicate.
The short answer is no, you should not treat the pairs from a pair end library as replicates.
The long answer, as the Dr. Matinez said, is that pairs doesn't represent the biological variation (it is the same extraction) or the technical variation (it is library preparation, the same run, and the same sequencing spot). If you want to have biological replicates you should have two independent RNA extraction of two different pools. If you want to have two different technical replicates you need to have two independent library preps. They can have different tags and can be run in the same lane, but still they should be independent preparation to capture the variability caused by the human manipulation.