unique-sam 0.2.4

Analyse sam file and keep the unique aligment record.

Unique-Sam

Unique-Sam is a simple command line tool to remove the duplicated alignments in the SAM file. If the MAPQ field of the alignment is available, unique-sam will keep one and only one alignment with the highest score. Otherwise, unique-sam will calculate a score according to the alignment’s MD or CIGAR field and use the calculated value to remove the duplicated alignments.

Install

• Install with the source code, in the source folder:

  python setup.py install

• If you have **pip** installed, you can simply run

  pip install unique-sam

  After installation you can access unique-sam from your command line.

Usage

unique-sam need a SAM format file to run properly. Before using unique-sam command, we must sort the sam file by the QNAMEfield. You can use samtools to achieve this purpose, refer to samtools for more helps:

samtools sort --help

For basic usage, in your command line environment:

unique-sam input.sam -o output.sam

-s Parameter

If you don’t have access to samtools, you can use -s option of unique-sam:

unique-sam -s input.sam -o output.sam

The sort functionality of unique-sam is implemented as

1. copy a temp file from the original sam file

2. extract the header of the sam file

3. sort the alignments with Bash sort program

-k Parameter

-k parameter give you the control on how to extract the alignment key from qname field the parameter of the -k is a regular expression. You should group the key part with parentheses. ** e.g. 1** > qname: N|GACGCGGATCTT/500407:4:H03E5AFXX:1:21109:5977:6969_2:N:0:ATACAA > > -k '(.*)\_[1-2](.*)' > > key will be: N|GACGCGGATCTT/500407:4:H03E5AFXX:1:21109:5977:6969:N:0:ATACAA > which will remove the _1/2 part of the qname.

** e.g. 2** > qname: HWI-ST667_0147:1:1101:1128:2079#CGATGT/1 > > -k '(.*)\/[1-2]' > > key will be: HWI-ST667_0147:1:1101:1128:2079#CGATGT > which will remove the /1/2 part of the qname.

For more about unique-sam run:

unique-sam --help

Unique Strategy

Following strategies are applied to find the unique & the best alignment

1. Keep the alignment pair that has the highest score. If more than one pairs are found to have the same “Highest Score”, these pairs will be removed.

2. Read1 and Read2 should be mapped on different strands.

3. The segment length decided by the read pairs should be longer than 0.7 * read length

Log File

All removed alignments will be written into log file input.sam.log under current folder. Each line of the log file start with a symbol and followed by the deleted alignment (the original alignment record in the input.sam). The symbol describe the reason of why this/these alignments should be removed. The specification of these symbols are listed in the follow table:

| Symbol | Description | | —— | ———– | | ! | Error lines | | < | Low score alignments | | = | Pairs with more than one best score | | ~ | Read pair mapped on the same strand | | ? | Segment length too short | | - | Invalid read1/2 information in flag filed or unmapped segment |

Copyright

Copyright (c) 2015 [e-mail]