#!/usr/bin/env python
# coding: utf-8

# ## Notebook for getting count matrix for individual crab RNAseq from crab transcriptome v 3.1

# #### Note: The individual crab RNAseq was not used in the assembly of crab transcriptome v 3.1      
# I will just be using samples for an individual crab over time: ambient and infected. Library IDs: 178, 359, and 463

# #### Notebook based on this from Steven Roberts: https://github.com/sr320/nb-2020/blob/master/C_bairdi/20-kallisto.ipynb

# In[1]:


get_ipython().system('/Applications/bioinfo/kallisto/kallisto')


# In[2]:


get_ipython().system('pwd')


# In[3]:


ls


# In[4]:


get_ipython().system('wget --no-check-certificate https://owl.fish.washington.edu/halfshell/genomic-databank/cbai_transcriptome_v3.1.fasta')


# In[5]:


get_ipython().system('/Applications/bioinfo/kallisto/kallisto  index -i cbai_transcriptome_31  cbai_transcriptome_v3.1.fasta')


# In[8]:


#sequence files for library 178 (single crab: ambient infected - day 0)
get_ipython().system('/Applications/bioinfo/kallisto/kallisto quant  -i /Users/graciecrandall/Documents/GitHub/paper-tanner-crab/notebooks/cbai_transcriptome_31  -t 4  -o kallisto/178_ambient_infected_0/  /Volumes/web/Atumefaciens/20200318_cbai_RNAseq_fastp_trimming/178_R1_001.fastp-trim.202003181815.fq.gz /Volumes/web/Atumefaciens/20200318_cbai_RNAseq_fastp_trimming/178_R2_001.fastp-trim.202003181815.fq.gz')


# In[9]:


#sequence files for library 359 (single crab: ambient infected - day 2)
get_ipython().system('/Applications/bioinfo/kallisto/kallisto quant  -i /Users/graciecrandall/Documents/GitHub/paper-tanner-crab/notebooks/cbai_transcriptome_31  -t 4  -o kallisto/359_ambient_infected_2/  /Volumes/web/Atumefaciens/20200318_cbai_RNAseq_fastp_trimming/359_R1_001.fastp-trim.202003182247.fq.gz /Volumes/web/Atumefaciens/20200318_cbai_RNAseq_fastp_trimming/359_R2_001.fastp-trim.202003182247.fq.gz')


# In[10]:


#sequence files for library 463 (single crab: ambient infected - day 17)
get_ipython().system('/Applications/bioinfo/kallisto/kallisto quant  -i /Users/graciecrandall/Documents/GitHub/paper-tanner-crab/notebooks/cbai_transcriptome_31  -t 4  -o kallisto/463_ambient_infected_17/  /Volumes/web/Atumefaciens/20200318_cbai_RNAseq_fastp_trimming/463_R1_001.fastp-trim.202003185732.fq.gz /Volumes/web/Atumefaciens/20200318_cbai_RNAseq_fastp_trimming/463_R2_001.fastp-trim.202003185732.fq.gz')


# ### Create matrix

# In[11]:


get_ipython().system('perl /Applications/bioinfo/trinityrnaseq-v2.8.6/util/abundance_estimates_to_matrix.pl')


# In[12]:


get_ipython().system('perl /Applications/bioinfo/trinityrnaseq-v2.8.6/util/abundance_estimates_to_matrix.pl  --est_method kallisto      --gene_trans_map none      --out_prefix /Users/graciecrandall/Documents/GitHub/paper-tanner-crab/analyses/kallisto-single_crab_over_time/kallisto-single-crab      --name_sample_by_basedir       /Users/graciecrandall/Documents/GitHub/paper-tanner-crab/notebooks/kallisto/178_ambient_infected_0/abundance.tsv       /Users/graciecrandall/Documents/GitHub/paper-tanner-crab/notebooks/kallisto/359_ambient_infected_2/abundance.tsv       /Users/graciecrandall/Documents/GitHub/paper-tanner-crab/notebooks/kallisto/463_ambient_infected_17/abundance.tsv')


# In[13]:


get_ipython().system('head /Users/graciecrandall/Documents/GitHub/paper-tanner-crab/analyses/kallisto-single_crab_over_time/kallisto-single-crab.isoform.counts.matrix')


# In[ ]: