MiXCR is a powerful computation program for TCR analysis. It can identify human TCR clones from raw sequences in a highly sensitive way. However, when the input sequences were not exclusively from TCR regions, some paralog sequences, such as those from immunoglobulin regions, could be mis-identified as TCR genes. Here we provide a bash script, named MixcrFilter, to filter out TCR clones which were falsely recognized by MiXCR from paralog sequences according to alignment results by BLAT.
The sample name and paths to the input files need to be customized by editing Lines 5-10 of MixcrFilter.sh. The required input files are:
- Raw RNA-seq reads in FASTQ format:
*.R1.fq.gz,*.R2.fq.gz - Reference genome file, perferable hg38, which has better annotation of TCR sequences:
hg38.fa
A sub-directary with the files with the sample name as the prefix, and following suffix:
| Suffix | Description |
|---|---|
| alignments.vdjca | Output from MiXCR align |
| clones.clns | Output from MiXCR assemble |
| clones.txt | Raw clones from MiXCR exportClones |
| cloneseq.fa | FASTA file of all nucleotide sequences of CDR3 region |
| cloneseq.psl | BLAT results of *.cloneseq.fa against reference genome |
| blat1.clones.txt | Clones excluding the ones which CDR3-containing reads mapped to non-TCR regions |
| blat2.clones.txt | Clones excluding the ones which CDR3 nt seq mapped to non-TCR regions |
| alignments.txt | Alignment data from MiXCR exportAlignments |
| alignedreads.fa | Original reads containing identified CDR3 sequences |
| alignedreads.psl | BLAT results of *.alignedreads.fa against reference genome |
| falseCDR3AA | A list of AA sequences of CDR3 from falsely mapped reads (Format: AA-seq, # In-target reads, # Out-target reads, In/Out Ratio) |
| flt.clones.txt | Filtered clones |