|
Q: Which alignment format is accepted by FusionAnalyser as input?
A: FusionAnalyser accepts both SAM and BAM/BAI as input.
Q: Which input format is the preferred one and why?
A: In many scenarios, BAM is preferred over SAM files. This is typically due to three reasons: binary files, such as the BAM, are processed faster; BAM is a compressed file format, so it’s more compact than SAM; thanks to its internal organization, BAM allows a fast retrieval of all the reads mapping to a specific interval. The identification of the same set of reads in a SAM files is much more time consuming. Not surprisingly, BAM is largely preferred over SAM in tools like read viewers, such as the IGV. Although in many contexts BAM is preferred over SAM, here SAM is the preferred file format. The reason is that, to correctly process a paired-end read, FusionAnalyser must analyze both reads in each pair. In the SAM format, read pairs are physically contiguous (which is not the case for BAM files) so virtually no computational time is required to search for the mate sequence.
Q: Does FusionAnalyser strictly requires the presence of the SAM header?
A: No, the presence of the header is not required.
Q: Where can I get a FusionAnalyser reference file?
A: Reference files can be downloaded here.
Q: How many reads/Gigabases do I need to sequence in order to run FusionAnalyser?
A: It is difficult to say. It heavily depends on the expression level of each fusion (the lower the expression, the higher the number of sequences required to identify fusion). Very good results are typically obtained with 5-6 Gigabases of paired-end sequencing.
Q: What’s the best way to train and familiarize with FusionAnalyser?
A: The best way is to use the training set.
Q: Does FusionAnalyser relies on a specific sequencing instrument/company?
A: No, as long as the alignment data are in SAM/BAM format.
Q: Does FusionAnalyser accept long sequencing reads?
A: Although FusionAnalyser was designed and tested using short-read sequencers, it can potentially be used also with (reasonably) long sequences.
Q: Which aligner is the preferred one to generate alignment data for FusionAnalyser?
A: BWA is the best choice. BWA is able to correctly process and align paired-end sequences when mates from a paired-end read align non-concordantly and to attribute the correct flag set. Due to a known issue, BOWTIE is not able to correctly process these data, however BOWTIE 2 should be (I didn’t test it yet). Any info/test done with BOWTIE 2 or other aligners is welcome!
|
