I would like to know how Trust4 can directly analyze paired-end .fastq format data from the 10X Genomics platform for single-cell analysis.？ #271
Comments
|
Here is the reply from the discussion just in case you missed it: Yes, you can. It would be something like running fastq files from this section: https://github.com/liulab-dfci/TRUST4?tab=readme-ov-file#10x-genomics-data-and-barcode-based-single-cell-data . For the running speed, which version of TRUST4 are you using? Which step do you find is too slow? |
|
I am currently using the Cell Ranger to analyze upstream FASTQ data to obtain BAM format data for 10X single-cell transcriptome analysis of the immune repertoire. Then, I use the command run-trust4 -t 25 -b /home/zxsys/data6/bam/SRR22007527_genome_bam.bam -f /home/zxsys/data6/hg38_bcrtcr.fa --ref /home/zxsys/data6/human_IMGT+C.fa --barcode CB to analyze the BAM data to obtain single-cell immune repertoire data. This workflow is too slow, preventing rapid completion of data analysis. I would now like to know how to use the Trust4 command to directly analyze single-cell transcriptome FASTQ data to obtain immune repertoire data, without first using Cell Ranger to analyze and obtain BAM. Currently, when I use the command run-trust4 -f hg38_bcrtcr.fa --ref human_IMGT+C.fa -u path_to_10X_fastqs/R2.fastq.gz --barcode path_to_10X_fastqs/R1.fastq.gz --readFormat bc:0:15 --barcodeWhitelist cellranger_folder/cellranger-cs/VERSION/lib/python/cellranger/barcodes/737K-august-2016.txt [other options] to analyze single-cell transcriptome data, it results in errors and the analysis cannot be completed. |
|
First of all, thank you for your reply. |
|
What error message did you get? Is your data 10X gene expression data or 10X vdj-kit data? Which version of TRUST4 are you using? Your command looks right to me. (Let's use this issue instead of the Discussion). |
|
Hello expert, I am currently using the following command which only supports single-end data. Could you provide a command for analyzing paired-end data? Since I am a beginner, there are many things I still need to learn. run-trust4 -f hg38_bcrtcr.fa --ref human_IMGT+C.fa -u path_to_10X_fastqs/R2.fastq.gz --barcode path_to_10X_fastqs/R1.fastq.gz --readFormat bc:0:15 --barcodeWhitelist cellranger_folder/cellranger-cs/VERSION/lib/python/cellranger/barcodes/737K-august-2016.txt [other options] |
|
This depends on your structure. For example, if the read is in both R1, R2, and barcode and UMI is also in R1's first 26bp (16bp barcode + 10bp UMI), You can use "-1 R1 -2 R2 --barcode R1 --readFromat bc:0:15,r1:26:-1" for this. |
|
Hello, I tested TRUST4 on raw FASTQ files derived from 10x, paired-end, 16bp barcode + 10bp UMI in R1's first 26bp, and ~40M reads, I used this code:
it ran well, but slowly, taking close to 4 hours: [Wed Oct 30 13:49:50 2024] TRUST4 v1.1.4-r534 begins. In comparison, cellranger vdj ran for half an hour, so, I would like to know is this normal or something wrong with me? |
|
It's normal. I think your data is from 10x VDJ kit. TRUST4 is designed for regular RNA-seq data, like regular 10x 5' scRNA-seq without VDJ amplification. The read coverage is sparse for untarget-amplified data, so the assembly procedure, including data preprocessing, is more complex to achieve good sensitivity. I'm working on improving the running efficiency for the targeted-amplified data, like TCR-seq/BCR-seq, in recent versions, but there is still a long way to go in the current code structure. |
|
Yes, the data was from 10x TCR/BCR Amplification Kit! And, thank you for your reply, looking forward to the new version! |
|
Hi, I'm trying to understand the output files generated by the command above, but have some questions, 1、_toassemble_1.fq,_toassemble_2.fq,*_toassemble_bc.fa files have the same number of reads, in my case is 20,173,744，can I regard it as the number of reads after barcode correction? but why more reads are found in the log file: [Mon Nov 18 12:03:24 2024] Read in and count kmers for 38800000 reads. 2、How can I find the number like 'Valid Barcodes' in 10x qc report, which means 'Fraction of reads with barcodes that match the whitelist after barcode correction.'? 3、Is there a detailed document about the output files? |
|
|
thanks for your reply! I used the following command:
the number of reads in the result_tmp_toassemble_bc.fa:
the nohup.out file:[Mon Nov 18 11:11:12 2024] TRUST4 v1.1.4-r534 begins. |
|
Hi, do you have any advice on this situation? Please let me know if you need more information. |
|
Since "result_tmp' is a pretty general name, I'm wondering whether you are running multiple instances of TRUST4 at the same time, and the intermediate files are overwritten by another sample? |
|
Thanks for your advice, I created a new directory, modified the '-o' parameter, executed the command only once and got the consistent result as before. the command: all files, and the number of reads in the toassemble_bc.fa: the nohup.out file: Is there any further testing I can do? |
|
Can you show me the first a few lines of the two input fastq files? |
R1_001.fastq.gz@LH00169:488:22HLWYLT4:7:1101:2141:1042 1:N:0:AAGATTGGAT+AAATCCCGCT R2_001.fastq.gz@LH00169:488:22HLWYLT4:7:1101:2141:1042 2:N:0:AAGATTGGAT+AAATCCCGCT |
|
Oh, the read count in the log will count the paired-end read twice, one for read1, and one for read2. So 20M read pairs would mean around 40M reads during the counting kmer. So the numbers are consistent. I should have given this explanation much earlier... |
|
That makes sense! Now we know,
So could we assume that,
|
|
|
Thanks very much for your reply! I will run more tests. |
I would like to ask how Trust4 can directly analyze paired-end .fastq format data from the 10X Genomics platform for single-cell analysis, instead of analyzing BAM format data. Can you provide support for this analysis? The current analysis speed is too slow.
run-trust4 -t 25 -b /home/zxsys/data6/bam/SRR22007527_genome_bam.bam -f /home/zxsys/data6/hg38_bcrtcr.fa --ref /home/zxsys/data6/human_IMGT+C.fa --barcode CB
Is it possible to directly use FASTQ format for paired-end single-cell data analysis without using BAM files, while still ensuring that Trust4 operates normally?
The text was updated successfully, but these errors were encountered: