Very low RNA splicing rate for pulmonary AT2 cells

Question

[e.r.zakiev]

I observe a very low mRNA splicing rate in AT2 cells (~6%).

I followed the very nice tutorial on RNA velocity analysis with scVelo by Sam Morabito and up until this point it went smoothly.

Important to note that

• This is the first time I'm trying the RNA velocity, so I might have dun goofed
• Multiple references in reputable journals discover RNA velocity in pulmonary tissues, but they never mention the splicing rate they observed
• I didn't use a masked gtf, because I had the counts from the CellRanger and the bam files already aligned to the Ensembl transcriptome/genome and it is unfeasible to re-align it to the UCSC genome which does provide a masked genome

But where do I even start digging?

Maybe it's because I have the loom files generated using the possorted_genome_bam.bam files which contain all the cells, but I do a lot of dead cell filtering in preparation of the sparse count matrix and the UMAP embedding from the classic scRNAseq counts using Seurat?

[GenoMax]

Can you add some real numbers? Just saying 6% gives no indication of how many genes you have been able to detect overall. Amount of reads that aligned etc.

[e.r.zakiev]

Hello GenoMax and thank you for your engagement! Where should I look this up?

adata

# AnnData object with n_obs × n_vars = 27639 × 19
#    obs: 'orig.ident', 'nCount_RNA', 'nFeature_RNA', 'percent.mt', 'cNMF_signature_k15_1', 'cNMF_signature_k15_2', 'cNMF_signature_k15_3', 'cNMF_signature_k15_4', 'cNMF_signature_k15_5', 'cNMF_signature_k15_6', 'cNMF_signature_k15_7', 'cNMF_signature_k15_8', 'cNMF_signature_k15_9', 'cNMF_signature_k15_10', 'cNMF_signature_k15_11', 'cNMF_signature_k15_12', 'cNMF_signature_k15_13', 'cNMF_signature_k15_14', 'cNMF_signature_k15_15', 'cNMF_signature_k8_1', 'cNMF_signature_k8_2', 'cNMF_signature_k8_3', 'cNMF_signature_k8_4', 'cNMF_signature_k8_5', 'cNMF_signature_k8_6', 'cNMF_signature_k8_7', 'cNMF_signature_k8_8', 'Schiebinger_MEF.identity', 'Schiebinger_Pluripotency', 'Schiebinger_Proliferation', 'Schiebinger_ER.stress', 'Schiebinger_Epithelial.identity', 'Schiebinger_ECM.rearrangement', 'Schiebinger_Apoptosis', 'Schiebinger_Senescence', 'Schiebinger_Neural.identity', 'Schiebinger_Trophoblast.identity', 'Schiebinger_X.reactivation', 'Schiebinger_XEN', 'Schiebinger_Trophoblast.progenitors', 'Schiebinger_Spiral.Artery.Trophpblast.Giant.Cells', 'Schiebinger_Spongiotrophoblasts', 'Schiebinger_Oligodendrocyte.precursor.cells.(OPC)', 'Schiebinger_Astrocytes', 'Schiebinger_Cortical.Neurons', 'Schiebinger_RadialGlia-Id3', 'Schiebinger_RadialGlia-Gdf10', 'Schiebinger_RadialGlia-Neurog2', 'Schiebinger_Long-term.MEFs', 'Schiebinger_Embryonic.mesenchyme', 'Schiebinger_Cxcl12.co-expressed', 'Schiebinger_Ifitm1.co-expressed', 'Schiebinger_Matn4.co-expressed', 'Schiebinger_2c', 'PanglaoDB_Airwayepithelialcells_Endoderm', 'PanglaoDB_Airwaygobletcells_Mesoderm', 'PanglaoDB_Alveolarmacrophages_Mesoderm', 'PanglaoDB_Ciliatedcells_Endoderm', 'PanglaoDB_Claracells_Endoderm', 'PanglaoDB_Ionocytes_Mesoderm', 'PanglaoDB_PulmonaryalveolartypeIcells_Endoderm', 'PanglaoDB_PulmonaryalveolartypeIIcells_Endoderm', 'CellsPositiveFor_cNMFsignature_k8_1', 'CellsPositiveFor_cNMFsignature_k8_2', 'CellsPositiveFor_cNMFsignature_k8_3', 'CellsPositiveFor_cNMFsignature_k8_4', 'CellsPositiveFor_cNMFsignature_k8_5', 'CellsPositiveFor_cNMFsignature_k8_6', 'CellsPositiveFor_cNMFsignature_k8_7', 'CellsPositiveFor_cNMFsignature_k8_8', 'CellsPositiveFor_cNMFsignature_k15_1', 'CellsPositiveFor_cNMFsignature_k15_2', 'CellsPositiveFor_cNMFsignature_k15_3', 'CellsPositiveFor_cNMFsignature_k15_4', 'CellsPositiveFor_cNMFsignature_k15_5', 'CellsPositiveFor_cNMFsignature_k15_6', 'CellsPositiveFor_cNMFsignature_k15_7', 'CellsPositiveFor_cNMFsignature_k15_8', 'CellsPositiveFor_cNMFsignature_k15_9', 'CellsPositiveFor_cNMFsignature_k15_10', 'CellsPositiveFor_cNMFsignature_k15_11', 'CellsPositiveFor_cNMFsignature_k15_12', 'CellsPositiveFor_cNMFsignature_k15_13', 'CellsPositiveFor_cNMFsignature_k15_14', 'CellsPositiveFor_cNMFsignature_k15_15', 'Clusterk8_maxscore', 'Clusterk15_maxscore', 'barcode', 'UMAP_1', 'UMAP_2', 'DIFFMAP_1', 'DIFFMAP_2', 'sample_batch', 'batch', 'initial_size_unspliced', 'initial_size_spliced', 'initial_size', 'n_counts'
#    var: 'Accession', 'Chromosome', 'End', 'Start', 'Strand'
#    uns: 'orig.ident_colors', 'log1p', 'neighbors'
#    obsm: 'X_pca', 'X_umap'
#    layers: 'ambiguous', 'matrix', 'spliced', 'unspliced', 'Ms', 'Mu'
#    obsp: 'distances', 'connectivities'

adata.layers['spliced']

# <27639x19 sparse matrix of type '<class 'numpy.float32'>'
#   with 161 stored elements in Compressed Sparse Row format>

adata.layers['unspliced']

# <27639x19 sparse matrix of type '<class 'numpy.float32'>'
#   with 11156 stored elements in Compressed Sparse Row format>

Here is the concatenated .loom data object that was spewed out but the velocyto run10x command:

ldata

# AnnData object with n_obs × n_vars = 59095 × 36601
#    obs: 'batch', 'initial_size_unspliced', 'initial_size_spliced', 'initial_size', 'sample_batch'
#    var: 'Accession', 'Chromosome', 'End', 'Start', 'Strand'
#   layers: 'ambiguous', 'matrix', 'spliced', 'unspliced'

ldata.layers['spliced']

# <59095x36601 sparse matrix of type '<class 'numpy.uint16'>'
#   with 4198501 stored elements in Compressed Sparse Row format>

ldata.layers['unspliced']

# <59095x36601 sparse matrix of type '<class 'numpy.uint16'>'
#   with 58874869 stored elements in Compressed Sparse Row format>

ldata.layers['matrix']

# <59095x36601 sparse matrix of type '<class 'numpy.float32'>'
#   with 62767930 stored elements in Compressed Sparse Row format>

[GenoMax]

Someone knowledgeable about this should be along to help. I asked you to add that info in anticipation since they would want to know some numbers instead of just a % value.

[LChart]

Is it possible that this is snRNA-seq data?

[e.r.zakiev]

Hmm. Shouldn't be, it's supposed to be a classical scRNAseq Single Cell 3' v3 10x assay with multiplexation of several samples in one plate. Encapsulation was performed for each sample separately.

Answer 1

[e.r.zakiev]

Hello people sorry I am dumb the problem lied in the fact that I had mouse cells but my .gtf transcriptome file was for the human. Now with the proper reference I have ~80% splicing rate, as expected

[e.r.zakiev]

For anyone it might be useful: a colleague of mine, who was not dumb to use the wrong annotation gtf as I did, also observed an important issue of difference in the chromosome annotation in gtf vs bam files. It is accompanied by the velocyto's warning:

WARNING - The .bam file refers to a chromosome ‘M+' not present in the annotation (.gtf) file

The gtf contained "chrM" for the mitochondrial chromosome while in the bam files it was denoted as "chrMT". After sed -ing the "chrM" into "chrMT" in the reference gtf file, the splicing rate has drastically improved for him (from ~15% to ~80%).