#SSW: A Python Wrapper for the SIMD Smith-Waterman #With Unicode Text Comparsion Function

Notice

This Repository is forked from vishnubob/ssw, a great implemenation of swith waterman with SSE SIMD Command. However due to the limation of internal structue, the original version is not able to handle to the needs of comparing more than 128 differenet symbols at once. That's why we ceate this repo.

How to test my code

I am not yet test the install command, it should work, but I am not sure. If you want to test it, please clone this repo, and run:

$python3 setup.py build_ext --inplace 

For compiling the c object

For testing (example), please run:

$python3 unissw.py data/test1.txt  data/test2.txt

Command Line parameters

-p: source data in pair format (ID1 \tab Text1 \tab ID2 \tab Text2 )

$python3 unissw.py -p data/pair2.tsv

-v: with varaint characters comparsion. It will read varaint character file from /data/varinats.txt. If two characters are varinats, it will get +2 (by default), rather than mismacth score (-3 by default)

$python3 unissw.py -pv data/pair1.tsv

-o: to save comapre result to an external file

python3 unissw.py -o a.txt -pv data/P185n1618_10000_IDText.tsv

-d: print debug message

python3 unissw.py -pvd data/P185n1618_10000_IDText.tsv

-c: change the path of configure file.

python3 unissw.py -c config.json -pvd data/P185n1618_10000_IDText.tsv

if you didn't specify by -c, the system will read config.json in package root by default.

Configre.json

I have move few impoartant setting out from my code. Now you can revise the configure file to set those parameters for your task. Now the possible settings are:

{
  "gap_open_penalty": -3, 
  "gap_extend_penalty": -2,
  "match_score": 3,
  "variant_match_score": 2,
  "mismacth_penalty": -3,
  "with_variant": "True"
  "variant_file": "data/variants.txt",
  "log_file": "batch-data/out/log.txt",
  "log_message_level": 3,
  "num_of_max_process": 1
  "common_char_count_th":-1
}

All of them are optional, and above value equals the defualt values in the unissw program.

log_message_level : 設定螢幕上與log檔案內的訊息層次，可能值為：

• 0: 不顯示log,
• 1: 顯示並列印整體工作進度,
• 2: 顯示並列印整體工作進度+列印各卷比較細節。
• 3: 顯示並列印整體工作進度+顯示並列印各卷比較細節 (Batch Mode Only)

common_char_count_th : 需搭配 config檔內，"data_type":sc 使用, 利用此值控制超過多少共用字的字串pair，需要進行 sw 比對。 (Batch Mode Only)

Batch Mode

run unissw_dila.py for batch mode process, it has two command line options:

-t : specify task file [required]. The task file is for setting details for multiple compare tasks.

python3 unissw_dila.py  -t batch-data/task.json

Details of task.json is in following section

-c: change the path of configure file. [optional]

python3 unissw_dila.py -c config.json -t batch-data/task.json

if you didn't specify by -c, the system will read config.json in package root by default.

This library supports mutiprocess now, just set "num_of_max_process" in the config file.

"common_char_count_th" is used to decide whether two sentences will be included in ssw comparsion or not. The paramater works only with batch mode with datatype sc, otherwise the paramter will be ingored.

task.json

In task.json, you can specify how to run multiple comapare tasks. Here is an example.

[
  {
    "task_id":1,    
    "data_folder": "/data/5_GitArea/ssw/batch-data",
    "data_type": "s",
    "sent_file1":"in/a-sentences.tsv",     /* required  */
    "sent_file2":"in/b-sentences.tsv",     /* required  */
    "pair_file": "in/sentence-pairs.tsv",  /* required  */
    "output_file":  "out/out-s-a-b.tsv"
  },
  {
    "data_folder": "/data/5_GitArea/ssw/batch-data",
    "data_type": "sc",
    "sent_file1":"in/sent-file1_withCCC.tsv",
    "sent_file2":"in/sent-file2_withCCC.tsv",
    "pair_file": "in/pair-file_withCCC.tsv",
    "output_file":  "out/out-s-a-b-CCC.tsv"
  },
  {
    "data_folder": "/data/5_GitArea/ssw/batch-data",
    "data_type": "r",
    "text_file1": "cbeta-txt/T0220/001.txt",
    "text_file2": "cbeta-txt/T0220/002.txt",
    "pair_file": "in/pair-file_withCCC.tsv",
    "output_file":  "out/out-s-a-b-CCC.tsv"
  },
  {
    "data_folder": "/data/5_GitArea/ssw/data",
    "data_type": "p",               
    "pair_file": "pair1.tsv",           /* required  */
    "output_file":  "/data/5_GitArea/ssw/data/out-t-1-2.tsv"
  },
  {
    "data_type": "t",
    "sent_file1":"/data/5_GitArea/ssw/data/1.txt",     /* required  */
    "sent_file2":"/data/5_GitArea/ssw/data/1.txt",     /* required  */
    "output_file":  "/data/5_GitArea/ssw/data/out-t-1-2.tsv"
  }
]

The example consists of three tasks of all three possible runing types with unissw and unissw dila. That are: data_type set to "s"(seperate mode),"sc" (seperate mode with common character count), "p" (pair mode), and "t" (two text modes).

Overview

SSW is a fast implementation of the Smith-Waterman algorithm, which uses the Single-Instruction Multiple-Data (SIMD) instructions to parallelize the algorithm at the CPU level. This repository wraps the SSW library into an easy to install, high-level python interface with no external library dependancies.

The SSW library is written by Mengyao Zhao and Wan-Ping Lee, and this python interface is maintained by Giles Hall.

Installation

To install the SSW python package, use pip:

$ pip install ssw

Example Usage

import ssw
aligner = ssw.Aligner()
alignment = aligner.align(reference="ACGTGAGAATTATGGCGCTGTGATT", query="ACGTGAGAATTATGCGCTGTGATT")
print(alignment.alignment_report())
Score = 45, Matches = 24, Mismatches = 0, Insertions = 0, Deletions = 1

ref   1   ACGTGAGAATTATGGCGCTGTGATT
          ||||||||||||| |||||||||||
query 1   ACGTGAGAATTAT-GCGCTGTGATT