biobakery / maaslin2 Goto Github PK
View Code? Open in Web Editor NEWMaAsLin2: Microbiome Multivariate Association with Linear Models
Home Page: http://huttenhower.sph.harvard.edu/maaslin2
License: Other
MaAsLin2: Microbiome Multivariate Association with Linear Models
Home Page: http://huttenhower.sph.harvard.edu/maaslin2
License: Other
Hi,
thank for your package.
I've just done a small test and had the error:
fit_data = Maaslin2(
Can you let me know how I can resolve this error before to use my own data.
Thank you,
Virgg
Hello,
It might be helpful to adjust the description of min_prevalence option from "The minimum percent of samples for which a feature is detected at minimum abundance" to "The minimum proportion (fraction) of samples for which a feature is detected at minimum abundance"
The way it is worded now sound like the default value "min_prevalence = 0.1 "is a percentage, but it's actually 10% not 0.1 percent.
Hi,
when installing with:
if(!requireNamespace("BiocManager", quietly = TRUE))
install.packages("BiocManager")
BiocManager::install("Maaslin2")
I get the following error:
ERROR: dependency 'lpsymphony' is not available for package 'Maaslin2'
R version 4.2.2
Thanks,
Theo
I was running the ready-made example in the function Maaslin2
help page:
input_data <- system.file(
'extdata','HMP2_taxonomy.tsv', package="Maaslin2")
input_metadata <-system.file(
'extdata','HMP2_metadata.tsv', package="Maaslin2")
fit_data <- Maaslin2(
input_data, input_metadata,'demo_output', transform = "AST",
fixed_effects = c('diagnosis', 'dysbiosisnonIBD','dysbiosisUC','dysbiosisCD', 'antibiotics', 'age'),
random_effects = c('site', 'subject'),
normalization = 'NONE',
reference = 'diagnosis,nonIBD',
standardize = FALSE)
This leads to the following error:
....
2023-04-28 10:41:54.896843 INFO::Writing heatmap of significant results to file: demo_output/heatmap.pdf
Error in xtfrm.data.frame(x) : cannot xtfrm data frames
In addition: Warning messages:
1: Model failed to converge with 1 negative eigenvalue: -5.6e+00
2: Model failed to converge with 1 negative eigenvalue: -1.1e+01
3: In checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, :
Model failed to converge with max|grad| = 0.00291214 (tol = 0.002, component 1)
4: Model failed to converge with 1 negative eigenvalue: -2.1e+02
5: Model failed to converge with 1 negative eigenvalue: -2.2e+02
Information my R session:
> sessionInfo()
R version 4.3.0 (2023-04-21)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 22.04.2 LTS
Matrix products: default
BLAS: /home/xxx/bin/R-4.3.0/lib/libRblas.so
LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.10.0
locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
[5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
time zone: Europe/Mariehamn
tzcode source: system (glibc)
attached base packages:
[1] stats4 stats graphics grDevices utils datasets methods
[8] base
other attached packages:
[1] doRNG_1.8.6 rngtools_1.5.2
[3] foreach_1.5.2 ANCOMBC_2.2.0
[5] lubridate_1.9.2 forcats_1.0.0
[7] stringr_1.5.0 dplyr_1.1.2
[9] purrr_1.0.1 readr_2.1.4
[11] tidyr_1.3.0 tibble_3.2.1
[13] ggplot2_3.4.2 tidyverse_2.0.0
[15] knitr_1.42 MicrobiomeStat_1.1
[17] Maaslin2_1.13.0 ALDEx2_1.32.0
[19] zCompositions_1.4.0-1 truncnorm_1.0-9
[21] NADA_1.6-1.1 survival_3.5-5
[23] MASS_7.3-59 tidySummarizedExperiment_1.10.0
[25] patchwork_1.1.2.9000 mia_1.8.0
[27] MultiAssayExperiment_1.26.0 TreeSummarizedExperiment_2.8.0
[29] Biostrings_2.68.0 XVector_0.40.0
[31] SingleCellExperiment_1.22.0 SummarizedExperiment_1.30.0
[33] Biobase_2.60.0 GenomicRanges_1.52.0
[35] GenomeInfoDb_1.36.0 IRanges_2.34.0
[37] S4Vectors_0.38.0 BiocGenerics_0.46.0
[39] MatrixGenerics_1.12.0 matrixStats_0.63.0
[41] BiocStyle_2.28.0 rebook_1.9.0
loaded via a namespace (and not attached):
[1] bitops_1.0-7 DirichletMultinomial_1.42.0
[3] doParallel_1.0.17 httr_1.4.5
[5] numDeriv_2016.8-1.1 backports_1.4.1
[7] tools_4.3.0 utf8_1.2.3
[9] R6_2.5.1 vegan_2.6-4
[11] lazyeval_0.2.2 mgcv_1.8-42
[13] rhdf5filters_1.12.0 permute_0.9-7
[15] withr_2.5.0 gridExtra_2.3
[17] cli_3.6.1.9000 logging_0.10-108
[19] biglm_0.9-2.1 sandwich_3.0-2
[21] mvtnorm_1.1-3 robustbase_0.95-1
[23] pbapply_1.7-0 proxy_0.4-27
[25] yulab.utils_0.0.6 foreign_0.8-84
[27] scater_1.28.0 decontam_1.20.0
[29] readxl_1.4.2 rstudioapi_0.14
[31] RSQLite_2.3.1 generics_0.1.3
[33] Matrix_1.5-4 biomformat_1.28.0
[35] ggbeeswarm_0.7.1 fansi_1.0.4
[37] DescTools_0.99.48 DECIPHER_2.28.0
[39] lifecycle_1.0.3 multcomp_1.4-23
[41] yaml_2.3.7 rhdf5_2.44.0
[43] grid_4.3.0 blob_1.2.4
[45] crayon_1.5.2 dir.expiry_1.8.0
[47] lattice_0.21-8 beachmat_2.16.0
[49] CodeDepends_0.6.5 pillar_1.9.0
[51] optparse_1.7.3 statip_0.2.3
[53] boot_1.3-28.1 gld_2.6.6
[55] estimability_1.4.1 codetools_0.2-19
[57] glue_1.6.2 data.table_1.14.8
[59] Rdpack_2.4 vctrs_0.6.2
[61] treeio_1.24.0 cellranger_1.1.0
[63] gtable_0.3.3 cachem_1.0.7
[65] xfun_0.39 rbibutils_2.2.13
[67] Rfast_2.0.7 coda_0.19-4
[69] pcaPP_2.0-3 modeest_2.4.0
[71] timeDate_4022.108 iterators_1.0.14
[73] statmod_1.5.0 gmp_0.7-1
[75] TH.data_1.1-2 ellipsis_0.3.2
[77] nlme_3.1-162 phyloseq_1.44.0
[79] bit64_4.0.5 filelock_1.0.2
[81] fBasics_4022.94 irlba_2.3.5.1
[83] vipor_0.4.5 rpart_4.1.19
[85] colorspace_2.1-0 DBI_1.1.3
[87] Hmisc_5.0-1 nnet_7.3-18
[89] ade4_1.7-22 Exact_3.2
[91] tidyselect_1.2.0 emmeans_1.8.5
[93] timeSeries_4021.105 bit_4.0.5
[95] compiler_4.3.0 graph_1.78.0
[97] htmlTable_2.4.1 BiocNeighbors_1.18.0
[99] expm_0.999-7 DelayedArray_0.25.0
[101] plotly_4.10.1 checkmate_2.2.0
[103] scales_1.2.1 DEoptimR_1.0-12
[105] spatial_7.3-16 digest_0.6.31
[107] minqa_1.2.5 rmarkdown_2.21.3
[109] base64enc_0.1-3 htmltools_0.5.5
[111] pkgconfig_2.0.3 lme4_1.1-33
[113] sparseMatrixStats_1.12.0 lpsymphony_1.28.0
[115] stabledist_0.7-1 fastmap_1.1.1
[117] rlang_1.1.0 htmlwidgets_1.6.2
[119] DelayedMatrixStats_1.22.0 energy_1.7-11
[121] zoo_1.8-12 jsonlite_1.8.4
[123] BiocParallel_1.34.0 BiocSingular_1.16.0
[125] RCurl_1.98-1.12 magrittr_2.0.3
[127] Formula_1.2-5 scuttle_1.10.0
[129] GenomeInfoDbData_1.2.10 Rhdf5lib_1.22.0
[131] munsell_0.5.0 Rcpp_1.0.10
[133] ape_5.7-1 viridis_0.6.2
[135] RcppZiggurat_0.1.6 CVXR_1.0-11
[137] stringi_1.7.12 rootSolve_1.8.2.3
[139] stable_1.1.6 zlibbioc_1.46.0
[141] plyr_1.8.8 parallel_4.3.0
[143] ggrepel_0.9.3 lmom_2.9
[145] splines_4.3.0 hash_2.2.6.2
[147] multtest_2.56.0 hms_1.1.3
[149] igraph_1.4.2 reshape2_1.4.4
[151] ScaledMatrix_1.7.1 rmutil_1.1.10
[153] XML_3.99-0.14 evaluate_0.20
[155] BiocManager_1.30.20 nloptr_2.0.3
[157] tzdb_0.3.0 getopt_1.20.3
[159] clue_0.3-64 rsvd_1.0.5
[161] xtable_1.8-4 Rmpfr_0.9-2
[163] e1071_1.7-13 tidytree_0.4.2
[165] viridisLite_0.4.1 class_7.3-21
[167] gsl_2.1-8 lmerTest_3.1-3
[169] memoise_2.0.1 beeswarm_0.4.0
[171] cluster_2.1.4 timechange_0.2.0
I asked this the forum and got no answer.
It is not clear where to find the "latest" version of Maaslin2. Installing from Github results in a v1.7.3, and installing from Bioconductor results in v1.10.0.
However the Bioconductor version is missing code that is > 9 months old according to the git blame (example: https://github.com/biobakery/Maaslin2/blame/master/R/Maaslin2.R#L974) .
This poses a problem when trying to update packages as it claims that Maaslin2 is outdated if you installed the Github repo. This would probably be fixed with simply making a new release on Github with a higher version than v1.10.0
The bioBakery support forum provides software support and tutorials for methods for microbial community profiling developed by the Huttenhower lab. Please consider creating a new topic in the bioBakery support forum before opening issues in Github.
It would be really nice if Maaslin2.log could include the version of Maaslin used as one of the lines in the log.
Hi,
Thanks for this wonderful tool.
I had an inquiry regarding the transformation implemented in the package.
For the CLR, the imputation is calculated as half the min feature for each sample. Why did not you consider using the robust CLR? Did you find in your testing any issues with it?
Thanks!
Thanks for a good tool.
I have four factors in a predictive variable; Q1, Q2, Q3, and Q4.
I've set Q1 as a reference.
However, I found that the order of the x-axis labels was Q1, Q3, Q2, and Q4 in scatter plots.
I tried to change the labels of the variable with A, B, C, and D, but the trouble was the same such as A, C, B, and D orders.
Could you check this situation?
The version of Maaslin2 I used is 1.8.0 in R.
Could you add scatter_first_n
(particularly given that heatmap_first_n
exists)?
Thanks,
Liam
Hi. I have a weighted population, which mean every person in my data may respresent different number of people. I wonder is it possible to use weight in Maaslin2? Like the "weights" parameter in the following command:
model1=glm(nodegree~treat,data=lalonde,family=binomial(),weights=iptw)
Hello all,
I was wondering if there are any options available for reducing or adjusting the font size in maaslin2 plots.
Thanks!
Hello MaAsLin2 Users,
Hi all,
Here is the error I get when I run maaslin2 in R. I was able to run it in Galaxy using pcl file, though. Don't know why says values are character.
fit_data = Maaslin2(pathabundance_relab, metadata_t, "results/2018-06-29-seq-QC-and-trimming/HUMANn2/pathabundance/MaAsLin")
[1] "Warning: Deleting existing log file: results/2018-06-29-seq-QC-and-trimming/HUMANn2/pathabundance/MaAsLin/maaslin2.log"
2019-07-22 18:43:54 INFO::Writing function arguments to log file
2019-07-22 18:43:54 INFO::Verifying options selected are valid
2019-07-22 18:43:54 INFO::Determining format of input files
2019-07-22 18:43:54 INFO::Input format is data samples as columns and metadata samples as columns
2019-07-22 18:43:55 INFO::Formula for fixed effects: expr ~ SampleID + FeedingType + DeliveryMode + Comments
2019-07-22 18:43:55 INFO::Running selected normalization method: TSS
Error in FUN(newX[, i], ...) : invalid 'type' (character) of argument
In addition: Warning message:
In vegan::decostand(features_norm, method = "total", MARGIN = 1, :
input data contains negative entries: result may be non-sense
Hi, I am trying to run Maaslin2 on Windows system using following command
Maaslin2(
input_data = otu.tab,
input_metadata = metadata,
output = 'C:\Users\lenovo\Desktop\tmp'
#transform = "AST",
fixed_effects = c('HBP','X1','PA', 'age', 'dietscore'),
random_effects = c('study'),
#normalization = 'NONE',
plot_heatmap = F,
plot_scatter = F,
standardize = FALSE)
Where it fails due to:
Error in file(file, ifelse(append, "a", "w")) :
cannot open the connection
In addition: Warning message:
In file(file, ifelse(append, "a", "w")) :
cannot open file './Scratch/tmp_1/maaslin2.log': No such file or directory
However, it works if I run it on linux system, so I am wandering whether Maaslin2 cannot run on Windows system
Hi,
I have been running Maaslin for a while and the values in my "value" column from the significant_results.tsv were always the different factors of the column. Recently, for some reason, the values are now changed to 1, 2, 3... instead. I don't know what the values 1, 2, 3 represent. I reran the code from before (where I got the actual names of my factors), however, am still getting this issue.
For example, this is an example of my code:
fit_data = Maaslin2(
input_data = input_data,
input_metadata = input_metadata,
normalization = "CSS",
standardize = FALSE ,
transform = "NONE",
analysis_method = "NEGBIN" ,
max_significance = 0.05,
output = "xxx",
fixed_effects = c("Sample"),
correction = "BH",
reference = c("Sample,aaa"),
min_abundance = 0,
min_prevalence = 0,
heatmap = TRUE,
plot_scatter = TRUE)
My four factors in Sample are aaa, bbb, ccc, ddd. aaa is my reference.
In the past, the "value" column of the significant_results.tsv would be:
value
bbb
bbb
ccc
ddd
bbb
ccc
Now, when I run Maaslin2, the value i get is:
value
2
2
3
4
2
3
Is there something I can change to return to bbb, ccc, ddd instead of 2, 3, 4?
I hope my explanation isn't confusing! Thank you for the great tool!
Carmen
maaslin2 is a great tool and we are planning to include maaslin2 into Galaxy and using it in comparative-analysis focused trainings.
There is already a wrapper for it, but it needs to be fixed and updated: https://github.com/galaxyproject/tools-iuc/tree/main/tools/maaslin2 - we will do that.
Could you maybe update the bioconda recipe with the newest release: https://github.com/bioconda/bioconda-recipes/blob/master/recipes/maaslin2/meta.yaml
So we can also bump the version in the Galaxy wrapper to provide the users with the newest version.
If you would like to be involved in the planned training or can provide specific scenarios (longitudinal, fixed-effects ... ) you would like to see in the training, please feel free to reach out as well.
Hi,
I'm having issues with the filtering by abundance when using different normalization methods. It seems Maaslin2
first runs the normalization of the data and then performs the filtering, however, it is hard to determine a number to set an abundance cut-off with normalized data. It would make more sense determine which features need to be filtered, normalize and then filter.
An example with the test data provided by Maaslin2
library(Maaslin2)
input_data <- system.file('extdata','HMP2_taxonomy.tsv', package="Maaslin2")
input_metadata <-system.file('extdata','HMP2_metadata.tsv', package="Maaslin2")
Model_1 <- Maaslin2(input_data,
input_metadata,
"/ebio/abt3_projects/small_projects/jdelacuesta/scratchpad",
normalization = "CLR",
fixed_effects = c('diagnosis', 'dysbiosisnonIBD','dysbiosisUC','dysbiosisCD', 'antibiotics', 'age'),
random_effects = c('site', 'subject'),
standardize = FALSE)
Model_2 <- Maaslin2(input_data,
input_metadata,
"/ebio/abt3_projects/small_projects/jdelacuesta/scratchpad",
normalization = "NONE",
fixed_effects = c('diagnosis', 'dysbiosisnonIBD','dysbiosisUC','dysbiosisCD', 'antibiotics', 'age'),
random_effects = c('site', 'subject'),
standardize = FALSE)
Using CLR transformation results in different number of filtered features:
From Model_1
2020-03-12 15:58:04 INFO::Writing function arguments to log file
2020-03-12 15:58:04 INFO::Verifying options selected are valid
2020-03-12 15:58:04 INFO::Determining format of input files
2020-03-12 15:58:04 INFO::Input format is data samples as rows and metadata samples as rows
2020-03-12 15:58:04 INFO::Formula for random effects: expr ~ (1 | site) + (1 | subject)
2020-03-12 15:58:04 INFO::Formula for fixed effects: expr ~ diagnosis + dysbiosisnonIBD + dysbiosisUC + dysbiosisCD + antibiotics + age
2020-03-12 15:58:04 INFO::Running selected normalization method: CLR
2020-03-12 15:58:04 INFO::Filter data based on min abundance and min prevalence
2020-03-12 15:58:04 INFO::Total samples in data: 1595
2020-03-12 15:58:04 INFO::Min samples required with min abundance for a feature not to be filtered: 159.500000
2020-03-12 15:58:04 INFO::Total filtered features: 51
From Model_2
2020-03-12 15:58:30 INFO::Writing function arguments to log file
2020-03-12 15:58:30 INFO::Verifying options selected are valid
2020-03-12 15:58:30 INFO::Determining format of input files
2020-03-12 15:58:30 INFO::Input format is data samples as rows and metadata samples as rows
2020-03-12 15:58:30 INFO::Formula for random effects: expr ~ (1 | site) + (1 | subject)
2020-03-12 15:58:30 INFO::Formula for fixed effects: expr ~ diagnosis + dysbiosisnonIBD + dysbiosisUC + dysbiosisCD + antibiotics + age
2020-03-12 15:58:30 INFO::Running selected normalization method: NONE
2020-03-12 15:58:30 INFO::Filter data based on min abundance and min prevalence
2020-03-12 15:58:30 INFO::Total samples in data: 1595
2020-03-12 15:58:30 INFO::Min samples required with min abundance for a feature not to be filtered: 159.500000
2020-03-12 15:58:30 INFO::Total filtered features: 0
Hi! I am working within TSD and to install pacakges I need to import the [.tar.gz] file into TSD and thereby install it within the "safe cloud". I downloaded it from here, https://bioconductor.org/packages/release/bioc/html/Maaslin2.html, but it is the 1.2.0 version. Can you share the most recent version.tar.gz with me? That would be much appreciated. Thanks,
Maria
Hello!
I am trying to run Maaslin2 with the code:
input_data = read.table(file = "4Masslin2_input.data_kos.taxonomy.archaea.mt.2group.tsv",
header = TRUE, sep = "\t")
rownames(input_data) <- input_data$Geneid_ord
input_data$Geneid_ord = NULL
metadata = read.table(file = "4Masslin2_metadata_kos.taxonomy.archaea.mt.2group.tsv",
header = TRUE, sep = "\t")
rownames(metadata) <- metadata$Geneid_ord
metadata$Geneid_ord = NULL
# Create the 'Ctrl' column
metadata$Ctrl <- ifelse(metadata$Diagnosis == "Ctrl", "Yes", "No")
# Create the 'PD' column
metadata$PD <- ifelse(metadata$Diagnosis == "PD", "Yes", "No")
# Create the 'iRBD' column
metadata$iRBD <- ifelse(metadata$Diagnosis == "iRBD", "Yes", "No")
reference <- unique(metadata$S)
reference <- c("Methanobrevibacter_A smithii","Methanobrevibacter_A smithii_A","Methanosphaera stadtmanae","Methanomethylophilus alvus","DTU008 sp001421185","Methanomassiliicoccus luminyensis","MX-02 sp006954405","Coprobacillus cateniformis","Methanobrevibacter_C arboriphilus_A","Methanosphaera cuniculi")
Maaslin2(input_data = input_data,
input_metadata = metadata,
fixed_effects = c("Ctrl", "PD", "iRBD", "S"),
reference = reference,
min_prevalence = 0,
output = "test",
transform = "LOG",
plot_heatmap = TRUE,
plot_scatter = TRUE,
heatmap_first_n = 50,
max_significance = 1)
Examples of my metadata and input data are below:
metadata
:
Diagnosis D P C O F G
K00053_1 Ctrl Archaea Methanobacteriota Methanobacteria Methanobacteriales Methanobacteriaceae Methanobrevibacter_A
K00053_2 Ctrl Archaea Methanobacteriota Methanobacteria Methanobacteriales Methanobacteriaceae Methanobrevibacter_A
K00053_3 Ctrl Archaea Methanobacteriota Methanobacteria Methanobacteriales Methanobacteriaceae Methanosphaera
K00053_4 Ctrl Archaea Thermoplasmatota Thermoplasmata Methanomassiliicoccales Methanomethylophilaceae Methanomethylophilus
K00053_5 PD Archaea Methanobacteriota Methanobacteria Methanobacteriales Methanobacteriaceae Methanobrevibacter_A
K00053_6 PD Archaea Methanobacteriota Methanobacteria Methanobacteriales Methanobacteriaceae Methanobrevibacter_A
S Ctrl PD iRBD
K00053_1 Methanobrevibacter_A smithii Yes No No
K00053_2 Methanobrevibacter_A smithii_A Yes No No
K00053_3 Methanosphaera stadtmanae Yes No No
K00053_4 Methanomethylophilus alvus Yes No No
K00053_5 Methanobrevibacter_A smithii No Yes No
K00053_6 Methanobrevibacter_A smithii_A No Yes No
input_data
:
tpm
K00053_1 166.502489
K00053_2 188.409788
K00053_3 69.970092
K00053_4 2.219452
K00053_5 642.522944
K00053_6 136.308126
As a result I receive an error:
2023-05-11 17:25:04 INFO::Writing function arguments to log file
2023-05-11 17:25:04 INFO::Verifying options selected are valid
2023-05-11 17:25:04 INFO::Determining format of input files
2023-05-11 17:25:04 INFO::Input format is data samples as rows and metadata samples as rows
2023-05-11 17:25:04 INFO::Formula for fixed effects: expr ~ Ctrl + PD + iRBD + S
Error in Maaslin2(input_data = input_data, input_metadata = metadata, :
Please provide the reference for the variable 'S' which includes more than 2 levels: Methanobrevibacter_A smithii, Methanobrevibacter_A smithii_A, Methanosphaera stadtmanae, Methanomethylophilus alvus, Methanomassiliicoccus_A intestinalis, UBA71 sp905187815, DTU008 sp001421185, Methanomassiliicoccus luminyensis, MX-02 sp006954405, Coprobacillus cateniformis, Methanobrevibacter_C arboriphilus_A, Methanosphaera cuniculi, Methanobrevibacter ruminantium_A.
Could you please suggest a solution to the error and probably the source of it?
Dear Maaslin2 developers,
using the function, everything looks fine and nice. Very useful indeed, Thanks!
I'm using far less data than the one used in your example data sets, and I would find extremely useful to add parameters to (optionally) map one of the variables in metadata to points colour or shape in scatterplots. It could help to better interpret the microbes/continuous data relation
Maybe also the possibility to plot label could be useful.
Best,
Francesco.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ 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. ๐๐๐
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 โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.