Illumina Sabela Andrea

1 Community B09

1.1 Full

b09 <- read.delim(
  "/home/miguel/Documentos/Miguel Díez Fernández de Bobadilla/b09_bis.tsv",
  col.names = c("qseqid", "sseqid", "pident", "length", "mismatch", "gapopen", 
                "qstart", "qend", "sstart", "send", "evalue", "bitscore", "stitle")
) %>%
  group_by(qseqid) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  rename(ID = stitle)

df2_sel <- df2 %>%
  select(ASV_ID3, B09_S1_L001, ID) %>%
  filter(B09_S1_L001 > 0) %>%
  mutate(sample = "B09_S1_L001") %>%
  left_join(b09, by = "ID") %>%
  group_by(ASV_ID3) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  mutate(
    qseqid = ifelse(
      is.na(qseqid),
      "NO",
      paste0("IS_", sub("([0-9]+).*", "\\1", qseqid))
    ),
    ASV_ID3 = paste(ASV_ID3, qseqid, sep = "|")
  )


ggplot(df2_sel, aes(x = factor(sample), y = B09_S1_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

1.2 Completness

ggplot(df2_sel, aes(x = factor(sample), y = B09_S1_L001, fill = qseqid)) +
  geom_col(position = "fill") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

1.3 Filter counts <2%

df2_sel <- df2_sel %>% filter(B09_S1_L001 >0.02) 

ggplot(df2_sel, aes(x = factor(sample), y = B09_S1_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

2 Community C09

2.1 Full

c09 <- read.delim(
  "/home/miguel/Documentos/Miguel Díez Fernández de Bobadilla/c09_bis.tsv",
  col.names = c("qseqid", "sseqid", "pident", "length", "mismatch", "gapopen", 
                "qstart", "qend", "sstart", "send", "evalue", "bitscore", "stitle")
) %>%
  group_by(qseqid) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  rename(ID = stitle)

df2_sel <- df2 %>%
  select(ASV_ID3, C09_S4_L001, ID) %>%
  filter(C09_S4_L001 > 0) %>%
  mutate(sample = "C09_S4_L001") %>%
  left_join(c09, by = "ID") %>%
  group_by(ASV_ID3) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  mutate(
    qseqid = ifelse(
      is.na(qseqid),
      "NO",
      paste0("IS_", sub("([0-9]+).*", "\\1", qseqid))
    ),
    ASV_ID3 = paste(ASV_ID3, qseqid, sep = "|")
  )

ggplot(df2_sel, aes(x = factor(sample), y = C09_S4_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

2.2 Completness

ggplot(df2_sel, aes(x = factor(sample), y = C09_S4_L001, fill = qseqid)) +
  geom_col(position = "fill") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

2.3 Filter counts <2%

df2_sel <- df2_sel %>% filter(C09_S4_L001 >0.02) 
df2_sel$sample <- "C09_S4_L001"

ggplot(df2_sel, aes(x = factor(sample), y = C09_S4_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

3 Community F05

3.1 Full

f05 <- read.delim(
  "/home/miguel/Documentos/Miguel Díez Fernández de Bobadilla/f05_bis.tsv",
  col.names = c("qseqid", "sseqid", "pident", "length", "mismatch", "gapopen", 
                "qstart", "qend", "sstart", "send", "evalue", "bitscore", "stitle")
) %>%
  group_by(qseqid) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  rename(ID = stitle)

df2_sel <- df2 %>%
  select(ASV_ID3, F05_S5_L001, ID) %>%
  filter(F05_S5_L001 > 0) %>%
  mutate(sample = "F05_S5_L001") %>%
  left_join(f05, by = "ID") %>%
  group_by(ASV_ID3) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  mutate(
    qseqid = ifelse(
      is.na(qseqid),
      "NO",
      paste0("IS_", sub("([0-9]+).*", "\\1", qseqid))
    ),
    ASV_ID3 = paste(ASV_ID3, qseqid, sep = "|")
  )



ggplot(df2_sel, aes(x = factor(sample), y = F05_S5_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

3.2 Completness

ggplot(df2_sel, aes(x = factor(sample), y = F05_S5_L001, fill = qseqid)) +
  geom_col(position = "fill") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

3.3 Filter counts <2%

df2_sel <- df2_sel  %>% filter(F05_S5_L001 >0.02) 
df2_sel$sample <- "F05_S5_L001"

ggplot(df2_sel, aes(x = factor(sample), y = F05_S5_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

4 Community E03

4.1 Full

e03 <- read.delim(
  "/home/miguel/Documentos/Miguel Díez Fernández de Bobadilla/e03_bis.tsv",
  col.names = c("qseqid", "sseqid", "pident", "length", "mismatch", "gapopen", 
                "qstart", "qend", "sstart", "send", "evalue", "bitscore", "stitle")
) %>%
  group_by(qseqid) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  rename(ID = stitle)

df2_sel <- df2 %>%
  select(ASV_ID3, E03_S6_L001, ID) %>%
  filter(E03_S6_L001 > 0) %>%
  mutate(sample = "E03_S6_L001") %>%
  left_join(e03, by = "ID") %>%
  group_by(ASV_ID3) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  mutate(
    qseqid = ifelse(
      is.na(qseqid),
      "NO",
      paste0("IS_", sub("([0-9]+).*", "\\1", qseqid))
    ),
    ASV_ID3 = paste(ASV_ID3, qseqid, sep = "|")
  )


ggplot(df2_sel, aes(x = factor(sample), y = E03_S6_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

4.2 Completness

ggplot(df2_sel, aes(x = factor(sample), y = E03_S6_L001, fill = qseqid)) +
  geom_col(position = "fill") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

4.3 Filter counts <2%

df2_sel <- df2_sel %>%  filter(E03_S6_L001 >0.02) 
df2_sel$sample <- "B09_S1_L001"

ggplot(df2_sel, aes(x = factor(sample), y = E03_S6_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

5 Community F06

5.1 Full

f06 <- read.delim(
  "/home/miguel/Documentos/Miguel Díez Fernández de Bobadilla/f06_bis.tsv",
  col.names = c("qseqid", "sseqid", "pident", "length", "mismatch", "gapopen", 
                "qstart", "qend", "sstart", "send", "evalue", "bitscore", "stitle")
) %>%
  group_by(qseqid) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  rename(ID = stitle)

df2_sel <- df2 %>%
  select(ASV_ID3, F06_S12_L001, ID) %>%
  filter(F06_S12_L001 > 0) %>%
  mutate(sample = "F06_S12_L001") %>%
  left_join(f06, by = "ID") %>%
  group_by(ASV_ID3) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  mutate(
    qseqid = ifelse(
      is.na(qseqid),
      "NO",
      paste0("IS_", sub("([0-9]+).*", "\\1", qseqid))
    ),
    ASV_ID3 = paste(ASV_ID3, qseqid, sep = "|")
  )



ggplot(df2_sel, aes(x = factor(sample), y = F06_S12_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

5.2 Completness

ggplot(df2_sel, aes(x = factor(sample), y = F06_S12_L001, fill = qseqid)) +
  geom_col(position = "fill") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

5.3 Filter counts <2%

df2_sel <- df2_sel %>% filter(F06_S12_L001 >0.02) 
df2_sel$sample <- "F06_S12_L001"

ggplot(df2_sel, aes(x = factor(sample), y = F06_S12_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

6 Community E06

6.1 Full

e06 <- read.delim(
  "/home/miguel/Documentos/Miguel Díez Fernández de Bobadilla/e06_bis.tsv",
  col.names = c("qseqid", "sseqid", "pident", "length", "mismatch", "gapopen", 
                "qstart", "qend", "sstart", "send", "evalue", "bitscore", "stitle")
) %>%
  group_by(qseqid) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  rename(ID = stitle)

df2_sel <- df2 %>%
  select(ASV_ID3, E06_S7_L001, ID) %>%
  filter(E06_S7_L001 > 0) %>%
  mutate(sample = "E06_S7_L001") %>%
  left_join(e06, by = "ID") %>%
  group_by(ASV_ID3) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  mutate(
    qseqid = ifelse(
      is.na(qseqid),
      "NO",
      paste0("IS_", sub("([0-9]+).*", "\\1", qseqid))
    ),
    ASV_ID3 = paste(ASV_ID3, qseqid, sep = "|")
  )


ggplot(df2_sel, aes(x = factor(sample), y = E06_S7_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

6.2 Completness

ggplot(df2_sel, aes(x = factor(sample), y = E06_S7_L001, fill = qseqid)) +
  geom_col(position = "fill") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

6.3 Filter counts <2%

df2_sel <- df2_sel %>% filter(E06_S7_L001 >0.02) 
df2_sel$sample <- "E06_S7_L001"

ggplot(df2_sel, aes(x = factor(sample), y = E06_S7_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

7 Community E07

7.1 Full

e07 <- read.delim(
  "/home/miguel/Documentos/Miguel Díez Fernández de Bobadilla/e07_bis.tsv",
  col.names = c("qseqid", "sseqid", "pident", "length", "mismatch", "gapopen", 
                "qstart", "qend", "sstart", "send", "evalue", "bitscore", "stitle")
) %>%
  group_by(qseqid) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  rename(ID = stitle)

df2_sel <- df2 %>%
  select(ASV_ID3, E07_S8_L001, ID) %>%
  filter(E07_S8_L001 > 0) %>%
  mutate(sample = "E07_S8_L001") %>%
  left_join(e07, by = "ID") %>%
  group_by(ASV_ID3) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  mutate(
    qseqid = ifelse(
      is.na(qseqid),
      "NO",
      paste0("IS_", sub("([0-9]+).*", "\\1", qseqid))
    ),
    ASV_ID3 = paste(ASV_ID3, qseqid, sep = "|")
  )


ggplot(df2_sel, aes(x = factor(sample), y = E07_S8_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

7.2 Completness

ggplot(df2_sel, aes(x = factor(sample), y = E07_S8_L001, fill = qseqid)) +
  geom_col(position = "fill") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

7.3 Filter counts <2%

df2_sel <- df2_sel %>% filter(E07_S8_L001 >0.02) 
df2_sel$sample <- "E07_S8_L001"

ggplot(df2_sel, aes(x = factor(sample), y = E07_S8_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

8 Community c11

8.1 Full

c11 <- read.delim(
  "/home/miguel/Documentos/Miguel Díez Fernández de Bobadilla/c11_bis.tsv",
  col.names = c("qseqid", "sseqid", "pident", "length", "mismatch", "gapopen", 
                "qstart", "qend", "sstart", "send", "evalue", "bitscore", "stitle")
) %>%
  group_by(qseqid) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  rename(ID = stitle)

df2_sel <- df2 %>%
  select(ASV_ID3, C11_S10_L001, ID) %>%
  filter(C11_S10_L001 > 0) %>%
  mutate(sample = "C11_S10_L001") %>%
  left_join(c11, by = "ID") %>%
  group_by(ASV_ID3) %>%
  slice_max(bitscore, n = 1) %>%
  ungroup() %>%
  mutate(
    qseqid = ifelse(
      is.na(qseqid),
      "NO",
      paste0("IS_", sub("([0-9]+).*", "\\1", qseqid))
    ),
    ASV_ID3 = paste(ASV_ID3, qseqid, sep = "|")
  )



ggplot(df2_sel, aes(x = factor(sample), y = C11_S10_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

8.2 Completness

ggplot(df2_sel, aes(x = factor(sample), y = C11_S10_L001, fill = qseqid)) +
  geom_col(position = "fill") +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )

8.3 Filter counts <2%

df2_sel <- df2_sel %>% filter(C11_S10_L001 >0.02) 
df2_sel$sample <- "C11_S10_L001"

ggplot(df2_sel, aes(x = factor(sample), y = C11_S10_L001, fill = ASV_ID3)) +
  geom_col(position = "fill") +
  scale_y_continuous(labels = percent_format()) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(angle = 90, vjust = 0.5, hjust = 1)
  ) +
  labs(
    x    = "Muestra",
    y    = "Porcentaje de lecturas",
    fill = "Taxon"
  )