about tumor and matched normal comparison

[Yuichi Shiraishi]

Hi,

I'm Yuichi Shiraishi, working at the Univ. Tokyo.

We are very interested in incorporating ruffus for our cancer genome sequence data analysis pipeline,

and now intensively evaluating it.

For cancer genome sequence analysis, the main purpose is to find somatic mutation 

by comparing tumor and normal DNAs from the same patients.

So the pipeline is 

1. the original data is sequence data (.fastq files) of several tumor and their matched controls.

(e.g., tumor samples: Tokyo1-T, Tokyo2-T, Tokyo3-T, their matched controls Tokyo1-N, Tokyo2-N, Tokyo3-N)

2. First, we want to align them to the reference genome

(e.g., tumor samples: Tokyo1-T.bam, Tokyo2-T.bam, Tokyo3-T.bam, their matched controls Tokyo1-N.bam, Tokyo2-N.bam, Tokyo3-N.bam)

3. Then, we get the somatic mutations by comparing within the same patients DNAs (e.g., Tokyo1-T.bam v.s. Tokyo1-N.bam -> Tokyo1.mutation.txt)

Also, the problem is even more difficult because;

1. Sometimes, several tumor samples share one controls, which in common for multi-region sequencing analysis.

(e.g., Tokyo1-N is the control for all Tokyo1-T, Tokyo2-T, Tokyo3-T)

2. We want to use all the controls (control panels) for accurate somatic mutation detection, and this strategy is adapted in many institute.

(e.g., Tokyo1-T v.s. Tokyo1-N, Tokyo2-N and Tokyo3-N will give us better results).

So our question is

1. 

We can obviously use @files decorator. But the manual says it is deprecated and should not be used.

http://www.ruffus.org.uk/tutorials/new_tutorial/deprecated_files.html

We actually tried @files decorator and it seemed to work.

Do you think can we continue to use @files ?

2. 

If we cannot use @files, then are there any approaches than can solve the above problems?

Thank you very much in advance.

Yuichi Shiraishi