Navigation Menu

Skip to content

Kingsford-Group/squid

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

92 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

SQUID logo{:height="50%" width="50%"}

OVERVIEW

SQUID is designed to detect both fusion-gene and non-fusion-gene transcriptomic structural variations from RNA-seq alignment.

SQUID paper is published at Genome Biology. To reproduce the result of applying SQUID on simulation data and previously studied cell lines, follow the instructions from squidtest

INSTALLING PRE-COMPILED BINARIES

You do NOT need to install SQUID before using it, find the binary release here!

BUILDING FROM SOURCE

You only need to build from source if either the pre-built binaries (see above) don't work on your system or you want to make a change to the SQUID code.

Compiling SQUID requires Boost, GLPK, BamTools. A step by step installation construction can be found here for linux, and here for mac.

On Mac, you need to additionly run the following command to dynamicly linking dependent libraries:

export DYLD_LIBRARY_PATH=<bamtools_folder>/lib
export DYLD_LIBRARY_PATH=<glpk_folder>/lib

USAGE

SQUID takes in a sorted BAM file of RNA-seq alignment and outputs the detection of TSVs. When the concordant and chimeric alignments are separated into two BAM files in the case of STAR alignment, the concordant BAM file must be sorted. The command to run SQUID and the parameters are as follows.

squid [options] -b <Input_sorted_BAM> -o <Output_Prefix>
Parameters Default value Data type Description
-c string
-f string
-pt 0 bool Phred type: 0 for Phred33, 1 for Phred64
-pl 10 int Maximum Length of continuous low Phred score to filter alignment
-pm 4 int Threshold to count as low Phred score
-mq 1 int Minimum mapping quality
-dp 50000 int Maximum paired-end aligning distance to be count as concordant alignment
-di 20 int Maximum distance of segment indexes to be count as read-through
-w 5 int Minimum edge weight
-r 8 double Discordant edge ratio multiplier (normal/tumor cell ratio)
-a 5 int Max allowed degree
-G 0 bool Whether or not output graph file (0 for not outputing, 1 for outputing)
-CO 0 bool Whether or not output ordering of connected components (0 for not outputing, 1 for outputing)
-TO 0 bool Whether or not output ordering of all segments (0 for not outputing, 1 for outputing)
-RG 0 bool Whether or not output rearranged genome sequence (0 for not outputing, 1 for outputing)

OUTPUT SPECIFICATION

  • <Output_Prefix>_sv.txt: a list of predicted TSV in bedpe format. This is the main output of SQUID. All positions in the file are 0-based. Each columns represents:
    • chr1: chromosome name of the first breakpoint.
    • start1: starting position of the segment of the first breakpoint, or the predicted breakpoint position if strand1 is "-".
    • end1: ending position of the segment of the first breakpoint, or the predicted breakpoint position if strand1 is "+".
    • chr2: chromosome name of the second breakpoint.
    • start2: starting position of the segment of the second breakpoint, or the predicted breakpoint position if strand2 is "-".
    • end2: ending position of the segment of the second breakpoint, or the predicted breakpoint position if strand2 is "+".
    • name: TSV is not named yet, this column shows with dot.
    • score: number of reads supporting this TSV (without weighted by Discordant edge ratio multiplier).
    • strand1: strand of the first segment in TSV.
    • strand2: strand of the second segment in TSV.
    • num_concordantfrag_bp1: number of concordant paired-end reads covering the first breakpoint. For a concordant paired-end read, it includes two ends and a inserted region in between, if any of the 3 regions covers the breakpoint, the read is counted in this number.
    • num_concordantfrag_bp2: number of concordant paired-end reads covering the second breakpoint. The count is defined in the same way as num_concordantfrag_bp1. example record:

    17 38135881 38136308 17 38137195 38137773 . 685 + + 106 221

    This means the right end (position 38136308) of segment 38135881-38136308 on chr17 is connected to the right end (position 38137773) of segment 38137195-38137773 also on chr17. The number of supporting reads for this TSV is 685. There are 106 concordant paired-end reads covering the first breakpoint (chr17 38136308), and 221 concordant paired-end reads covering the second breakpoint (chr17 38137773).

    5 176370330 176370489 8 128043988 128044089 . 328 - + 588 1029

    This means the left end (position 176370330) of segment 176370330-176370489 on chr5 is connected with the right end (position 128044089) of segment 128043988-128044089 on chr8. There are 588 concordant reads covering the first breakpoint (chr5 176370330), and 1029 concordant reads covering the second breakpoint (chr8 128044089).
  • <Output_Prefix>_graph.txt: genome segment graph, will be output only if -G is set to 1. It has two types of records, nodes (or segment) and edges.
    • node: for each node, the following information are included. ID, start position, end position, label for connected component.
    • edge: for each edge, the following information are included. ID, node id for the first segment, strand of the first segment, node id for the second segment, strand for the second segment, edge weight.
  • <Output_Prefix>_component_pri.txt: ordering of each connected component by ILP, will be output only if -CO is set to 1. In this file, ordering of each connected component will be output into a line. Each segment is represented by its id, with a "-" in the front if the ordering suggests the segment should be using its reverse strand.
  • <Output_Prefix>_component.txt: ordering of the entire genome segments, will be output only is -TO is set to 1. In this file, each newly generated chromosome will be output into one line. Other spefications are the same as <Output_Prefix>_component_pri.txt.
  • <Output_Prefix>_genome.fa: sequence of rearranged genome sequence, where each chromosome corresponds to one line in <Output_Prefix>_component.txt.

EXAMPLE WORKFLOW

Suppose you have the alignment BAM file, and chimeric BAM file generated by STAR (https://github.com/alexdobin/STAR), run SQUID with:

squid -b alignment.bam -c chimeric.bam -o squidout

Or a combined BAM file of both concordant and discordant alignments generated by BWA (http://bio-bwa.sourceforge.net/) or SpeedSeq (https://github.com/hall-lab/speedseq), run SQUID with

squid --bwa -b combined_alignment.bam -o squidout

An example can be run be downloading the sample data (sampledata.tgz) from (https://cmu.box.com/s/e9u6alp73rfdhfve2a51p6v391vweodq) into example folder, and decompress it with

tar -xzvf sampledata.tgz

Run SQUID command in example/SQUIDcommand.sh. Or if you want to test the workflow of STAR and SQUID, make sure STAR is in your path, and run bash script example/STARnSQUIDcommand.sh.

cd example
./SQUIDcommand.sh
./STARnSQUIDcommand.sh

Annotate SQUID output

To label the predicted TSVs as fusion-gene or non-fusion-gene type, and retrieve the corresponding gene names of fusion-gene TSVs, you can use the following python script.

Python dependencies:

  • numpy

Usage:

python <squid_folder>/utils/AnnotateSQUIDOutput.py [options] <GTFfile> <SquidPrediction> <OutputFile>

Note that the GTF file must have the same chromosome name as in SQUID output, and must contain 3 attributes in the transcript record: transcript ID, gene ID, and gene symbol (or gene name).

Options Default value Data type Description
--geneid gene_id string GTF gene ID attribute string, the attribute name in GTF record that corresponds to the gene ID
--genesymbol gene_name string GTF gene symbol attribute string, the attribute name in GTF record that corresponds to the gene symbol