search

JialiUMassWengLab / TEMP

TEMP is a software package for detecting transposable elements (TEs) insertions and excisions from pooled high-throughput sequencing data
GNU General Public License v2.0
21 stars 19 forks source link

BWA version #4

Open nelson42 opened 10 years ago

nelson42 commented 10 years ago

Hi, I have noticed some issues when trying to run your software. It seems that you rely on using the old version of bwa (pre. 0.7) to run the pipeline and for creating the bam input for the system. This seems to be because you use the old XT:A: bam flags to determine pairs of interest. These flags are not included when using the preferred "bwa mem" mode in version 0.7 onwards. In addition there seems to be a known bug in these versions of bwa when using aln and sampe. This means that some samples will fail for what seems to be no known reason (this bug is over a year old so seems unlikely bwa are going to correct it). Do you have plans to modify TEMP to run on the current version of bwa, using the mem method, as I am working on a TE detection pipeline that runs several different TE detection methods and would need this change to include your method. Thanks very much, Mike

JialiUMassWengLab commented 10 years ago

Hi Mike, Thanks for your comments. I'll try to accommodate the "bwa mem" mapping method.

ghannum commented 10 years ago

Hi,

Any update on the status of this request? I too am interested in running TEMP with the newest bwa.

Thanks, Greg

JialiUMassWengLab commented 9 years ago

Hey guys,

I made the modifications now that TEMP is compatible with BWA mem.

I take advantage of two tags in the SAM files outputted by BWA mem: "AS:i:" which is the alignment score and "XS:i:" which is the alternative alignment score.

Now users have the option to specify a score difference between those two scores by using option "-x". All reads with difference between the primary alignment and alternative alignment exceeds the user specified number will be considered "uniquely mapped".

I would suggest the value "-x 30" as the initial attempt. In BWA mem default scoring scheme, a mismatch is equivalent to a 5 point penalty. Therefore "-x 30" is equivalent of saying if the second best alignment has >= 6 more mismatches than the best alignment of a read then we consider it a "uniquely mapped" read. Of course you guys can try other values to fit your specific purposes.

Please let me know if you encounter any problems with this new modifications.

Best, Jiali