library type specification UN

nghiavtr / FuSeq

GNU General Public License v3.0

32 stars 12 forks source link

library type specification UN #2

Closed gudeqing closed 5 years ago

gudeqing commented 5 years ago

hi， It seems that library type specification is not correct as I got Exception: [unknown library format string : UN]

nghiavtr commented 5 years ago

Hello, Thank you for using FuSeq.

It is a bit strange because "UN" is the default setting of FuSeq. Can you give more details the issue you got?

Best, Nghia

gudeqing commented 5 years ago

hi, Nghia, I can excute the following comand. I guess IU means unstranded lib instead, am I right ? And '-l' is a mandatory argument !

FuSeq -o quasiMap -l IU --biasCorrect -i TxIndexer_idx -1 $1 -2 $2 -p 8 -g Homo_sapiens.GRCh37.75.gtf

nghiavtr commented 5 years ago

I see, so you concern about the library type at this step.

You are right, IU means unstranded library in the mapping process and it is a mandatory argument, and should not be changed.

The control of library type for fusion detection is set in the params.txt, please edit the parameters in that file when necessary,

Best, Nghia

gudeqing commented 5 years ago

Yes, thank you for your reply, and Hope you can update your document on this issue because I think library type is important at quasimap step, if possible, library-type should be detected automatically.
By the way, I appreciate your wonderful work, it is fast and easy to install and easy to use. Yet, I hope one sample may help you to improve FuSeq, the sample "MCF-7" is from https://chimpipe.readthedocs.io/en/latest/tutorial.html . For this sample, chimpipe and star-fusion both can detect 5 validated fusion genes, while FuSeq can only detect 3 validated ones.

Best, gudeqing

nghiavtr commented 5 years ago

Hi Gudeqing,

Many thanks for your comments and suggestions to improve FuSeq. We will consider the improvements for both documentation and codes in the next version of FuSeq The sample is interesting and under investigation.

Best, Nghia

nghiavtr commented 5 years ago

Btw, I have a quick check the validated fusions from your link for the sample and discovered that one of the reasons that FuSeq cannot detect the other fusions because two of them are too close (less than 100 000 bases (default of FuSeq) for RPS6KB1-VMP1 (74 936) and SMARCA4-CARM1 (81 642)).

When revising two parameters in the params.txt (minGeneDist= 50000 and minJunctionDist= 50000), one fusion is added (RPS6KB1-VMP1) in the results of FuSeq. The method still cannot detect SMARCA4-CARM1, I believe it depends on the suitable parameter settings for FuSeq, but not have time for investigation.

Best, Nghia