tanaylab / mcatac Goto Github PK
View Code? Open in Web Editor NEWMetacell analysis for ATAC data
Home Page: https://tanaylab.github.io/mcATAC/
License: Other
Metacell analysis for ATAC data
Home Page: https://tanaylab.github.io/mcATAC/
License: Other
This should work (mcmd$st is a cell-type annotation):
ann_list = generate_pheatmap_annotation(clust_vec = mcmd$st, feature_type = "cell_type", feature_annotation = "color")
But gives an error:
`
Error in stop_vctrs(class = c(class, "vctrs_error_names"), ...): Names must be unique.
✖ These names are duplicated:
:=
(!!feature_annotation, name)) # at line 67-68 of file /home/feshap/src/mcATAC/R/utils.R:=
(!!feature_annotation, name)):=
(!!feature_annotation, name))> devtools::load_all("~/src/mcATAC/")
ℹ Loading mcATAC
ℹ Parallelization enabled. Using 77 threads.
> gset_genome("mm10")
> mcc <- mcc_read("/net/mraid14/export/tgdata/users/yonshap/proj/embenh/data/mcc_lg/")
✔ Succesfully read a ScCounts object from /net/mraid14/export/tgdata/users/yonshap/proj/embenh/data/mcc_lg/
✔ Succesfully read a McCounts object from /net/mraid14/export/tgdata/users/yonshap/proj/embenh/data/mcc_lg/
> mcc_to_tracks(mc_counts=mcc, track_prefix='test_gastru_large_tracks', overwrite = T, normalize = T)
→ Normalizing each metacell by its total counts
Error in do.ply(i) :
task 1 failed - "error in evaluating the argument 'x' in selecting a method for function 't': Cholmod error 'problem too large' at file ../Core/cholmod_dense.c, line 102"
This is the slots of the object:
@mat
: a numeric matrix where rows are peaks and columns are metacells.
Can be a sparse matrix.
@peaks
: a misha intervals set with the peak definitions.
@genome
: genome assembly of the peaks
@egc
: a numeric matrix which contains normalized metacell accessibility.
@fp
: a matrix showing for each peak (row) the relative enrichment of umis
in log2 scale.
@metadata
: a tibble with a column called 'metacell' and additional
metacell annotations.
But after saving it to h5ad we don't have the egc and fp. From the code:
slots <- slots[slots %!in% c("egc", "fp", "mat", "peaks", "genome", "metadata", "ignore_peaks", "ignore_pmat", "rna_egc", "tad_based")]
Now that we manipulate the matrix for normalization we probably want to save all of them to re-load them, both in R and in python. We should just use the general layers of the anndata to do this
if a valid peaks interval set is passed (i.e. has 'chrom','start','end'
), but lacking peak_name
column - call peak_names()
internally to solve this issue, instead of aborting:
peak_set = fread('path/to/peak/file.txt')
mcatac_obj <- mcc_to_mcatac(mcc, peaks = peak_set, metadata = mc_md)
Error in `mcc_to_mcatac()`:
! The `peaks` must have a column called peak_name
Solution:
if(!has_name(peak_set)){ peak_set$peak_name = peak_names(peak_set) }
(or vice versa)
I think they should be 1-based. But it should at least be consistent...
> wd <- "/net/mraid14/export/tgdata/users/yonshap/proj/mmcortex/"
> setwd(wd)
> library(metacell)
> scdb_init('scdb')
initializing scdb to scdb
> devtools::load_all("/home/feshap/src/mcATAC")
ℹ Loading mcATAC
ℹ Parallelization enabled. Using 77 threads.
> mca <- mcATAC::import_from_h5ad(file = '/net/mraid14/export/tgdata/users/yonshap/proj/mmcortex/data/test_anndata_atac_metacell.h5ad', genome = 'mm10')
• Reading /net/mraid14/export/tgdata/users/yonshap/proj/mmcortex/data/test_anndata_atac_metacell.h5ad
! h5ad file doesn't have the mc_size_eps_q at the uns section. Using the default: 0.1
→ No id was given, setting id to Jennifer_Carter
• Setting egc cell size to 11312.1833139651 (the 0.1 quantile of metacell sizes)
✔ Successfully loaded an `McATAC` object with 454 metacells and 98905 ATAC peaks
> mc_rna <- scdb_mc('pl_cort')
> mca <- add_mc_rna(atac_mc = mca, mc_rna = mc_rna)
Warning messages:
1: mc_rna contains 1 metacells not present in the McATAC object: "454"
2: McATAC object contains 1 metacells not present in mc_rna: "0"
>
Suggestions:
small comment.
when re-creating tracks to the same directory and passing overwrite=True, it complains about an already-existing marginal track.
setting create_marginal_track=False, still results in an Error message:
"Error: File .../marginal already exists"
Run this:
devtools::load_all("~/src/mcATAC")
wd <- "/net/mraid14/export/tgdata/users/yonshap/proj/mmcortex"
setwd(wd)
mcmd <- readr::read_tsv('./BonevCollab/mcmd_pl_cort.tsv')
mcc <- mcc_read('./data/mmcortex_mcc/', id = 'mmcortex')
pks_all <- readRDS('./data/all_peaks_for_mmcortex_atac_mc.rds')
atac_mc <- mcc_to_mcatac(mc_counts = mcc, peaks = pks_all, metadata = dplyr::rename(mcmd, metacell = mc, cell_type = st))
Output:
> devtools::load_all("~/src/mcATAC")
Loading mcATAC
ℹ Parallelization enabled. Using 77 threads.
> wd <- "/net/mraid14/export/tgdata/users/yonshap/proj/mmcortex"
> setwd(wd)
> mcmd <- readr::read_tsv('./BonevCollab/mcmd_pl_cort.tsv')
Rows: 454 Columns: 11
── Column specification ───────────────────────────────────────────────────────────────────────────────────[...]
> mcc <- mcc_read('./data/mmcortex_mcc/', id = 'mmcortex')
|--------------------------------------------------|
[...]
=======|
✔ Succesfully read a McCounts object from ./data/mmcortex_mcc/
> pks_all <- readRDS('./data/all_peaks_for_mmcortex_atac_mc.rds')
> atac_mc <- mcc_to_mcatac(mc_counts = mcc, peaks = pks_all, metadata = dplyr::rename(mcmd, metacell = mc, cell_type = st))
97%...100%
81%...100%
94%...100%
87%...100%
96%...100%
59%...75%...100%
100%
Error in do.ply(i) :
task 1 failed - "error in evaluating the argument 'x' in selecting a method for function 't': error in evaluating the argument 'x' in selecting a method for function 't': contrasts can be applied only to factors with 2 or more levels"
>
In plot_tracks_at_locus, if you choose to plot an interval and not a gene TSS extended left and right, you lose the option to plot gene expression. There should be an option to plot a specified interval (like a TAD) and still show gene expression on the side.
I'm not sure why but when I built an atac_mc object using mcc_to_mcatac
many functions (such as: rna_atac_cor_knn
, plot_atac_atac_cor
, etc...) doesn't handle the input properly..
example:
my_mc_atac <- readRDS('/home/tomgo/raid/proj/mtec_populations/analysis_dir/multiome_mc_model/2022/atac_analysis/objs/mc_atac_obj.Rds')
plot_atac_atac_cor(atac_mc = my_mc_atac)
Error: "x" argument must be a matrix of numeric values
This breaks filter_features
(and its tests):
atac_sc <- import_from_10x("pbmc_data", genome = "hg38", id = "pbmc", description = "PBMC from a healthy donor - granulocytes removed through cell sorting (10k)")
atac_sc <- filter_features(atac_sc, minimal_max_umi = 3, min_peak_length = 200, max_peak_length = 1000, max_peak_density = 250)
#> Error in `name_enhancers()` at mcATAC/R/PeakIntervals.R:86:12:
#> ! Class of `atac` is not recognized (should be either ScATAC, McATAC or PeakIntervals
> write_sc_counts_from_fragments("./raid/proj/matching/data/pbmc_granulocyte_sorted_10k_atac_fragments.tsv", out_dir="./raid/proj/matching/data/pbmc_fragment_reads", genome = "hg38", overwrite=T, cell_na→ Writing to ./raid/proj/matching/data/pbmc_fragment_reads
ℹ 'tabix' was not found or an index file doesn't exist. Using all chromosomes
ℹ Processing 504 genomic bins of maximal size 50000000
→ Processing "chr1_0_50000000"
→ Processing "chr1_50000000_100000000"
→ Processing "chr1_100000000_150000000"
→ Processing "chr1_150000000_200000000"
→ Processing "chr1_200000000_248956422"
→ Processing "chr1_GL383518v1_alt_0_182439"
→ Processing "chr1_GL383519v1_alt_0_110268"
→ Processing "chr1_GL383520v2_alt_0_366580"
→ Processing "chr1_KI270706v1_random_0_175055"
→ Processing "chr1_KI270707v1_random_0_32032"
→ Processing "chr1_KI270708v1_random_0_127682"
gzip: ./raid/proj/matching/data/pbmc_granulocyte_sorted_10k_atac_fragments.tsv: not in gzip format
→ Processing "chr1_KI270709v1_random_0_66860"
gzip: ./raid/proj/matching/data/pbmc_granulocyte_sorted_10k_atac_fragments.tsv: not in gzip format
→ Processing "chr1_KI270710v1_random_0_40176"
gzip: ./raid/proj/matching/data/pbmc_granulocyte_sorted_10k_atac_fragments.tsv: not in gzip format
...
...
✔ Finished processing "chrUn_KI270753v1_0_62944"
✔ Finished processing "chrX_0_50000000"
✔ Finished processing "chrUn_KI270756v1_0_79590"
✔ Finished processing "chrUn_KI270752v1_0_27745"
✔ Finished processing "chrX_KI270913v1_alt_0_274009"
✔ Finished processing "chrUn_KI270754v1_0_40191"
✔ Finished processing "chrX_KI270880v1_alt_0_284869"
✔ Finished processing "chrUn_KI270751v1_0_150742"
✔ Finished processing "chrUn_KI270757v1_0_71251"
✔ Finished processing "chrY_0_50000000"
✔ Finished processing "chrX_KI270881v1_alt_0_144206"
✔ Finished processing "chrX_50000000_100000000"
✔ Finished processing "chrX_100000000_150000000"
✔ Finished processing "chrY_KI270740v1_random_0_37240"
✔ Finished processing "chrY_50000000_57227415"
✔ Finished processing "chrX_150000000_156040895"
ℹ Created sparse matrices for 504 genomic bins
✔ Created an ScCounts object at ./raid/proj/matching/data/pbmc_fragment_reads
> scc <- scc_read("./raid/proj/matching/data/pbmc_fragment_reads/")
Error in do.ply(i) :
task 1 failed - "skip=2 but the input only has 2 lines"
In addition: Warning messages:
1: Problem with `mutate()` column `start`.
ℹ `start = as.numeric(start)`.
ℹ NAs introduced by coercion
2: The following bins were not found in the genome: "chr1_GL383518v1_alt_0_182439", "chr1_GL383519v1_alt_0_110268", "chr1_GL383520v2_alt_0_366580", "chr1_KI270706v1_random_0_175055",
"chr1_KI270707v1_random_0_32032", "chr1_KI270708v1_random_0_127682", "chr1_KI270709v1_random_0_66860", "chr1_KI270710v1_random_0_40176", "chr1_KI270711v1_random_0_42210",
"chr1_KI270712v1_random_0_176043", "chr1_KI270713v1_random_0_40745", "chr1_KI270714v1_random_0_41717", "chr1_KI270759v1_alt_0_425601", "chr1_KI270760v1_alt_0_109528", "chr1_KI270761v1_alt_0_165834",
"chr1_KI270762v1_alt_0_354444", "chr1_KI270763v1_alt_0_911658", "chr1_KI270764v1_alt_0_50258", "chr1_KI270765v1_alt_0_185285", "chr1_KI270766v1_alt_0_256271", "chr1_KI270892v1_alt_0_162212",
"chr10_GL383545v1_alt_0_179254", "chr10_GL383546v1_alt_0_309802", "chr10_KI270824v1_alt_0_181496", "chr10_KI270825v1_alt_0_188315", "chr11_GL383547v1_alt_0_154407", "chr11_JH159136v1_alt_0_200998",
"chr11_JH159137v1_alt_0_191409", "chr11_KI270721v1_random_0_100316", "chr11_KI270826v1_alt_0_186169", "chr11_KI270827v1_alt_0_67707", "chr11_KI270829v1_alt_0_204059", "chr11_KI270830v1_alt_0_177092",
"chr11_KI270831v1_alt_0_296895", "chr11_KI270832v1_alt_0_210133", "chr11_KI270902v1_alt_0_106711", "chr11_KI270903v1_alt_0_214625", "chr11_KI270927v1_alt_0_218612", "chr12_GL383549v1_alt_0_120804",
"chr12_GL383550v2_alt_0_169178", "chr12_GL383551v1_alt_0_184319", "chr12_GL383552v1_alt_0_138655", "chr12_GL383553v2_alt_0_152874", "chr12_GL877875v1_alt_0_167313", "chr12_GL877876v1_alt_0_408271",
"chr12_KI270833v1_alt_0_76061", "chr12_KI270834v1_alt_0_119498", "chr12_KI270835v1_alt_0_238139", "chr12_KI270836v1_alt_0_56134", "chr12_KI270837v1_alt_0_40090", "chr12_KI270904v1_alt_0_572349",
"chr13_KI270838v1_alt_0_306913", "chr13_KI270839v1_alt_0_180306", "chr13_KI270840v1_alt_0_191684", "chr13_KI270841v1_alt_0_169134", "chr13_KI270842v1_alt_0_37287", "chr13_KI270843v1_alt_0_103832",
"chr14_GL000009v2_random_0_201709", "chr14_GL000194v1_random_0_191469", "chr14_GL000225v1_random_0_211173", "chr14_KI270722v1_random_0_194050", "chr14_KI270723v1_random_0_38115",
"chr14_KI270724v1_random_0_39555", "chr14_KI270725v1_random_0_172810", "chr14_KI270726v1_random_0_43739", "chr14_KI270844v1_alt_0_322166", "chr14_KI270845v1_alt_0_180703",
"chr14_KI270846v1_alt_0_1351393", "chr14_KI270847v1_alt_0_1511111", "chr15_GL383554v1_alt_0_296527", "chr15_GL383555v2_alt_0_388773", "chr15_KI270727v1_random_0_448248", "chr15_KI270848v1_alt_0_327382",
"chr15_KI270849v1_alt_0_244917", "chr15_KI270850v1_alt_0_430880", "chr15_KI270851v1_alt_0_263054", "chr15_KI270852v1_alt_0_478999", "chr15_KI270905v1_alt_0_5161414", "chr15_KI270906v1_alt_0_196384",
"chr16_GL383556v1_alt_0_192462", "chr16_GL383557v1_alt_0_89672", "chr16_KI270728v1_random_0_1872759", "chr16_KI270853v1_alt_0_2659700", "chr16_KI270854v1_alt_0_134193", "chr16_KI270855v1_alt_0_232857",
"chr16_KI270856v1_alt_0_63982", "chr17_GL000205v2_random_0_185591", "chr17_GL000258v2_alt_0_1821992", "chr17_GL383563v3_alt_0_375691", "chr17_GL383564v2_alt_0_133151", "chr17_GL383565v1_alt_0_223995",
"chr17_GL383566v1_alt_0_90219", "chr17_JH159146v1_alt_0_278131", "chr17_JH159147v1_alt_0_70345", "chr17_JH159148v1_alt_0_88070", "chr17_KI270729v1_random_0_280839", "chr17_KI270730v1_random_0_112551",
"chr17_KI270857v1_alt_0_2877074", "chr17_KI270858v1_alt_0_235827", "chr17_KI270859v1_alt_0_108763", …. This usually means that the genome ("hg38") is not compatible with the bins.
library(mcATAC)
wd <- '/net/mraid14/export/tgdata/users/yonshap/proj/mmcortex'
setwd(wd)
my_genome <- "mm10"
gset_genome(my_genome)
mca <- readRDS('./output/mcatac/mmcortex_mcatac_feat_peaks.rds')
scc <- scc_read(path = './data/frag_reads_28122022/')
✔ Succesfully read a ScCounts object from ./data/frag_reads_28122022/
> scpeaks <- scc_to_peaks(sc_counts = scc, peaks = dplyr::select(mca@peaks, chrom, start, end, peak_name))
Error in data.frame(i = all_0_idxs, j = 1, x = 0) :
arguments imply differing number of rows: 0, 1
> devtools::load_all("~/src/mcATAC/")
ℹ Parallelization enabled. Using 77 threads.
> setwd('/net/mraid14/export/tgdata/users/yonshap/proj/mmcortex')
> my_genome <- "mm10"
> gset_genome(my_genome)
> options(gmax.data.size = 1e+9)
> mctracks <- readRDS('./data/mmcortex_mctracks.rds')
> exons <- gintervals.load('intervs.global.exon')
> gene <- 'Satb2'
> gene_exons <- dplyr::filter(exons, geneSymbol == gene)
> gene_intervals <- dplyr::summarise(gene_exons, chrom = unique(chrom), start = min(start), end = max(end))
> mct_plot_region(mct = mctracks, intervals = gene_intervals, downsample_n = 1e+5)
→ Extracting region chr1:56793980-56971340
Error: Failed to allocate shared memory: Cannot allocate memory
Memory usage of the library can be controlled via gmax.data.size option (see options, getOptions).
> setwd("/net/mraid14/export/tgdata/users/yonshap/proj/mmcortex")
> devtools::load_all("~/src/mcATAC")
ℹ Loading mcATAC
ℹ Parallelization enabled. Using 77 threads.
> gset_genome("mm10")
> mcc <- mcc_read('./data/frag_read_mcc/')
> set_parallel(20)
ℹ Parallelization enabled. Using 20 threads.
> mcc_to_tracks(mcc, "mmcortex_microcluster", create_marginal_track=FALSE, overwrite=T)
ℹ Creating tracks for 192 metacells
ℹ Smoothing over 201 bp window
ℹ Tracks resolution: 10 bp
→ Extracting per-metacell data
→ Creating mmcortex_microcluster.mc1 track
→ Creating mmcortex_microcluster.mc2 track
→ Creating mmcortex_microcluster.mc3 track
→ Creating mmcortex_microcluster.mc4 track
→ Creating mmcortex_microcluster.mc5 track
…
77%...→ Creating mmcortex_microcluster.mc180 track
72%...72%...77%...90%...77%...100%
→ Creating mmcortex_microcluster.mc173 track
90%...90%...100%
100%
4%...13%...4%...22%...13%...31%...22%...40%...31%...50%...40%...59%...50%...63%...59%...72%...63%...77%...72%...90%...100%
77%...90%...100%
Error in do.ply(i) :
task 21 failed - "Problem with `filter()` input `..1`.
ℹ Input `..1` is `metacell == "21"`.
✖ cannot allocate vector of size 5.1 Gb"
In addition: Warning message:
In mclapply(argsList, FUN, mc.preschedule = preschedule, mc.set.seed = set.seed, :
scheduled cores 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 14, 15, 16, 17, 19 did not deliver results, all values of the jobs will be affected
different functions under feature_selection.R don't accept mcATAC objects as input, and it's not clear what is the correct input for them.
(e.g., get_peaks_coverage_stats / plot_peak_coverage_distribution ,etc...)
documentation says it should be "an ScATAC or McATAC object", but mcATAC is raising error
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.