missing values in 'row.names' when running transcript quantification

[ChangqingW]

Thank you for your reply. However, I am encountering this issue in the released version. I can confirm that the data and reference genome are correct — I have already validated them using the Python implementation of FLAMES — and I need a solution for the R version.

[M::worker_pipeline::6584.84731.81] mapped 638043 sequences
[M::worker_pipeline::6611.35731.77] mapped 638621 sequences
[M::worker_pipeline::6628.681*31.68] mapped 415399 sequences
[M::main] Version: 2.30-r1287
[M::main] CMD: /home/user3/miniconda3/envs/Minimap2/bin/minimap2 --eqx -N 100 -ax map-ont -y -t 80 /home/user3/scRNAseq/data/Nanopore/combind/transcript_assembly.fa /home/user3/scRNAseq/data/Nanopore/combind/DSE_matched_reads_dedup.fastq
[M::main] Real time: 6629.722 sec; CPU: 210019.095 sec; Peak RSS: 12.038 GB
Sorting by CB for oarfish quantifaction
Sorting BAM file with 80 threads...
[bam_sort_core] merging from 8 files and 80 in-memory blocks...
Running minimap2 with args: --eqx -N 100 -ax map-ont -y -t 80

Generating transcript count matrix

2025-08-28T23:20:09.544843Z INFO oarfish: setting user-provided filter parameters.
2025-08-28T23:20:09.761978Z INFO oarfish::alignment_parser: read header from BAM file /home/user3/scRNAseq/data/Nanopore/combind/DT.small_realign2transcript.bam, contains 295,302 reference sequences.
2025-08-28T23:20:09.762002Z INFO oarfish::alignment_parser: saw minimap2 as a program in the header; proceeding.
2025-08-28T23:20:09.762005Z INFO oarfish: calculating seqcol digest
2025-08-28T23:20:09.845052Z INFO oarfish: done calculating seqcol digest
2025-08-28T23:20:09.866604Z INFO oarfish: parsed reference information for 295,302 transcripts.
Error in DataFrame(x = seq_len(ncol(a1)), row.names = nms) :
missing values in 'row.names'
8.
stop("missing values in 'row.names'")
7.
DataFrame(x = seq_len(ncol(a1)), row.names = nms)
6.
SummarizedExperiment(...)
5.
SingleCellExperiment::SingleCellExperiment(assays = list(counts = mtx))
4.
parse_oarfish_sc_output(oarfish_out, annotation, outdir)
3.
quantify_transcript_oarfish(annotation, outdir, config, pipeline,
...)
2.
quantify_transcript(annotation = annotation, outdir = outdir,
config = config, pipeline = "sc_multi_sample", samples = names(fastqs))
1.
FLAMES::sc_long_multisample_pipeline(annotation = annotation,
fastqs = fastqs, outdir = outdir, genome_fa = genome_fa,
minimap2 = minimap2_bin, k8 = k8_bin, barcodes_file = barcodes_file,
expect_cell_numbers = 10000)

Originally posted by @haili2717096-lab in #62

[ChangqingW]

FLAMES/R/quantification.R

Line 335 in 273f0f6

rownames(mtx) <- read.delim(paste0(oarfish_out, ".features.txt"), header = FALSE)$V1

The row names are read from oarfish's [sample_name].features.txt and none of them are supposed to be missing. Could you run ls -hl in the output folder and post the output? If there are .features.txt files, could you run head on DT.small.features.txt and post the output? @haili2717096-lab

[haili2717096-lab]

Of course — I ran head DT.small.features.txt.
head DT.small.features.txt
ENSSSCG00000048769_792_3867_1
ENSSSCT00000107943
ENSSSCT00000078085
ENSSSCT00000107944
ENSSSCG00000037372_14879_19171_1
ENSSSCG00000037372_14879_19171_2
ENSSSCG00000037372_15142_22152_1
ENSSSCT00000046027
ENSSSCG00000027257_23676_40032_1
ENSSSCG00000027257_23784_27847_1

[haili2717096-lab]

sum(is.na(mtx))
[1] 0
sum(mtx == "")
[1] 0
sum(trimws(mtx) == "")
[1] 0
which(is.na(mtx) | trimws(mtx) == "")
integer(0)
class(mtx)
[1] "character"
length(mtx)
[1] 295302
object.size(mtx)
30041984 bytes

[ChangqingW]

That's odd. The file exists and the content seems good. We could try debugging the function in R.

debug(FLAMES:::parse_oarfish_sc_output)
oarfish_out <- "/path/to/head DT.small" # path to the DT.small.features.txt file but delete the .features.txt  part
annotation <- annotation # the one you supplied to sc_long_multisample_pipeline
outdir <- outdir
x <- FLAMES:::parse_oarfish_sc_output(oarfish_out, annotation, outdir)

This should get you into R's debug interface, you can hit n and enter to run the next line. After running the first three lines, could you check the following:

class(mtx)
head(rownames(mtx))
sum(is.na(rownames(mtx)))

class(mtx)
[1] "character"

This is very odd, it is meant to be a dgTMatrix from Matrix package. Can you also run head DT.small.count.mtx?