{:height="50%" width="50%"}
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
You do NOT need to install SQUID before using it, find the binary release here!
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
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) |
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
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:
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 |