ERROR: Unknown option: --no-home

[TinyTasy]

Dear Bambu Team,

Thank you for your lovely tool! I was trying out bambu-singlecell-spatial with the following code on my single-cell PacBio Kinnex data :

nextflow run GoekeLab/bambu-singlecell-spatial --bams /path/to/aligned.bam
--genome /path/to/fasta.fa
--annotation /path/to/annotation.gtf
--chemistry 10x3v3 --ncore 4 --outdir /path/to/output -with-singularity lingminhao/bambusc:beta1.2

When I run this, I get the following error:

Command exit status:
1

Command output:

Command error:
ERROR: Unknown option: --no-home

My nextflow version is 24.10.5 from bioconda and my singularity version is 2.4.2 from bioconda, too.

Any help would be greatly appreciated!

Best wishes,
Anastasiya

[lingminhao]

Hi @TinyTasy ,

This pipeline was developed under the nextflow version 23.04.3 and thus it might not be compatible with the later version. Can you try running again by adding NXF_VER=23.04.3 before your command line nextflow run ... so that it can run with a specified version. Let me know if it works.

Also, the singularity version can be a bit outdated, thus it would be better if you can update the singularity version to, say, 3.6.3 as well, even though I think that this might not be an issue.

Best Regards,
Min Hao.

[TinyTasy]

Dear @lingminhao,

Thanks a lot for your answer! Changing the nextflow version to 23.04.3 and singularity to 3.6.3 let me overcome the -no--home error. I also had to unset $XDG_RUNTIME_DIR.

Bambu managed to start then, but now I encounter a new error:

Error: BiocParallel errors
1 remote errors, element index: 1
0 unevaluated and other errors
first remote error:
Error in bambu.readsByFile(bam.file = reads[i], genomeSequence = genomeSequence, : No reads left after filtering.
Execution halted
INFO: Cleaning up image...

Could this be because of the tags in PacBio's bam files? I am starting from the aligned bam file which I aligned with pbmm2. I believe that the barcodes are stored in the CB:Z: tag and the UMIs in another tag. Is there a way on how I can run bambu with the aligned bam files of PacBio?

Sadly, I cannot let it run with the raw fastq outputs, as the raw output of PacBio is the unaligned BAM format. Sorry if I missed something, but I really appreciate your help!

Best,
Anastasiya

[lingminhao]

Hi @TinyTasy, it is totally okay to start running this pipeline from the bam file directly. There are three ways we extract the barcode and UMI.

1. directly from the read name but it has to be in this format: ATCCGTCCAACGGGTA_TTGCTGGCGTGT#46f0ce76-6a12-4a12-a707-2ffef7be7594_+1of1, where ATCCGTCCAACGGGTA is the cell barcode, TTGCTGGCGTGT is the UMI and 46f0ce76-6a12-4a12-a707-2ffef7be7594_+1of1 is the read name
2. from the tag in bam file. But it has to be in the BC tag for cell barcode and UG tag for the UMI
3. you can provide a .tsv or .csv file (with no header) that contains the read name in the bam file, the cell barcode, and the UMI barcode to the --barcode_map argument in nextflow run . You can find an example here

If all of the above does not work, can you send us a very tiny subsample of bam file, with the command you ran with, so I can further diagnose the problem?

Best Regards,
Min Hao

[TinyTasy]

Dear @lingminhao ,

Thanks a lot for your help! I have now created a .csv barcode_map, nevertheless, I sadly still get this error:

--- Start generating read class files ---
Removing 189675522 reads that were not assigned barcodes. If this is unexpected check the barcode map input
Primary alignment selection time 0.001 mins.
Only 0 detected. Not performing UMI deduplication. If this is unexpected, double check the --chemistry argument

My barcode map looks somewhat like this:

molecule_144941121,TAGACTGACCGTCGCA,CCATACAATATT
molecule_146268118,TAGCTGAACAGCTATC,CGGATTAGTCGA
molecule_73054370,CGATCTCACCAGACCC,TGGGGGCAGTCG
molecule_143026613,TACCTTTGAAGTGTGG,GCTACCACGAAG

I would appreciate it a lot if I could send you a very tiny subsample of my files! Should I send you that via mail?

Thank you so much for your help!

Best,
Anastasiya

[lingminhao]

Hi @TinyTasy, sure! You may send to my organisation email: lingmh@gis.a-star.edu.sg .

Do provide me with the command line you used to run the nextflow pipeline as well. Thanks!

[TinyTasy]

Thanks a lot @lingminhao, I am in the process of writing you the mail!

[lingminhao]

Hi @TinyTasy , I have a check with the files you provided, and realised that the read name format in the bam file does not match with the ones from your barcode_map.csv.gz. In particular, the read name format in the bam file is molecule/num but the ones from barcode_map.csv.gz is molecule_num.

Can you update the format and try again to see if it works now?

Best Regards,
Min Hao

[TinyTasy]

Hello @lingminhao,

Thank you for pointing out the different naming convention! I have changed the molecule number to the same names, but I am still getting the same error. I also checked for the permissions, but it all seems to be fine. Does it run on your side when you change the naming convention?

I'm sorry for the inconveniences, thank you so much for your patience and help so far!

Best,
Anastasiya

[lingminhao]

Hi @TinyTasy ,

I have fixed the bug. Can you try running again the command line ? Note that you have to:

1. use --barcodeMap instead of --barcode_map
2. remove the --chemistry argument as you don't do barcode demultiplexing
3. you might need to remove the pulled nextflow pipeline in ~/.nextflow/assets in order to use the updated pipeline.

I tried running the pipeline again at my end using a filtered barcode_map, and the error you mentioned disappear. However, you might still not able to finish the run as it introduces another error. Very likely due to the fact that the dataset you provided is too small.

I suggest you can try running it again using larger test dataset (like around 10K reads ..?). If that works, then you can run on full dataset. I also suggest to use at least 16 cores since you had over hundred millions of reads.

It's great to have you testing the tool, and we hope to make it a lot better in the near future :) So, don't hesitate to ask.

Best Regards,
Min Hao

[TinyTasy]

Dear @lingminhao,

Thanks again so much for your help. Just as you suggested, I used the --barcodeMap argument, removed the --chemistry argument and removed the .nextflow/assets. Sadly, I have only 4 cores available, so I cannot use more cores.

When I used it on only hundred cells, it did not work, as you predicted.
I used it on my full dataset then (~30 000 cells, ~ 260 M reads), but now I get the following error:

14:55:06 Writing NN index file to temp file /tmp/Rtmp35ezp8/filef7033f55
Unable to open: No such file or directory (2)
14:55:06 Searching Annoy index using 1 thread, search_k = 3000
Unable to open: No such file or directory (2)
Error in if (fsize > 2147483647) { :
missing value where TRUE/FALSE needed
Calls: clusterCells ... find_nn -> annoy_nn -> annoy_search -> annoy_search_parallel
Execution halted

On the Seurat github, I have seen people having a similar error. I'm really sorry for inconveniencing you further, but can you please help me here?

Thank you again so much :)

Best wishes,
Anastasiya

[TinyTasy]

Dear @lingminhao,

I am not quite sure why exactly, but after re-running bambu on my data, it finally worked and I got all the output needed. Thank you again so much for your help, I appreciate it a lot!

Now, I only have one more question: I am a bit unsure on how to get a Seurat object out of the output. The importBambu() function does not work for me. I see that as an output I have the combined__counts_gene.mtx and combined___counts_transcript.mtx and I suppose that these are the count matrices I need for a Seurat object. However, I am unsure of the barcodes and the genes (for the creation of the Seurat object) - do I just use the output that isoseq gave me for the barcodes?

I apologize for the inconveniences and thanks again so much for your help, it is greatly appreciated!

Best wishes,
Anastasiya

[lingminhao]

Dear @TinyTasy ,

sorry for the late reply, recently has been quite busy and I am glad to see you managed to run it in the end. Out of curiosity, may I know how long you took to run 30k cells, 260M reads with 4 cores (I presume)? This will be an useful statistics for us to improve the tools later on.

By the way, to answer your question, you should have two .tsv fils with something like xxx_txANDgenes.tsv and xxx_sampleData.tsv. The first tsv contains the row names and the second tsv contains the column names of the count matrices, which you can supply to create the Seurat object.

Thanks for testing !
Best Regards,
Min Hao

[bpuzek]

Dear @lingminhao ,

Thank you for the cool package!

I am trying out the multisample mode of bambu-clump and am encountering an error:

Error in bambu.readsByFile(bam.file = reads[i], genomeSequence = genomeSequence, : No reads left after filtering.

Running the same sample by providing --bams .bam and --barcodeMap in single-sample mode works fine.

Could it be that the bug fix you mentioned above was applied just for single-sample and not multi-sample runs?

I’ve sent first couple of lines of .bam and barcode map files and my samplesheet via email.

Many thanks in advance,
Barbara