scClassifier2 0.0.10

a tool for single cell data

scClassifier2

Deep bag-of-genes model for single cell classification

Citation

Bin Zhou, Fan Yang, Ting Chen, Feng Zeng. Seeing cells as bags-of-genes: a simple deep generative model for platform-independent single cell classification. Submission. 2021

Installation

1. Install pytorch according to your computational platform
2. Install dependencies: pip3 install numpy scipy pandas scikit-learn pyro-ppl matplotlib

Prepare matrix and label files

1. scClassifier2 accepts as input the log-transformed gene matrix in the MatrixMarket format usually end in ".mtx", where rows are cells and columns are genes.
2. The label file can be either the CSV format or the TSV format, one label per line.
3. Data gives some examples of matrix and label files.

Tutorial

Please refer to the PBMC68k example.

Usage

usage: scClassifier2.py [-h] [--cuda] [--jit] [-n NUM_EPOCHS] [--aux-loss] [-alm AUX_LOSS_MULTIPLIER] [-enum ENUM_DISCRETE]
                        [--sup-data-file SUP_DATA_FILE] [--sup-label-file SUP_LABEL_FILE] [--unsup-data-file UNSUP_DATA_FILE]
                        [--unsup-label-file UNSUP_LABEL_FILE] [-64] [-lt] [--cross-validation-fold CROSS_VALIDATION_FOLD] [-zd Z_DIM]
                        [-hl HIDDEN_LAYERS [HIDDEN_LAYERS ...]] [-lr LEARNING_RATE] [-dr DECAY_RATE] [-de DECAY_EPOCHS] [-b1 BETA_1]
                        [-bs BATCH_SIZE] [-rt] [-log LOGFILE] [--seed SEED] [--best-valid-model BEST_VALID_MODEL] [--best-aux-model BEST_AUX_MODEL]